Prashant Bharam
Göteborg (Gothenburg), Sweden Full work rights, no sponsorship needed PMP certified Program + Product AI-driven delivery

I deliver programs and ship products in regulated fintech, and I build the AI behind them.

Startup speed. Banking discipline. Technical execution.

Senior Program Manager and Product Manager with nine years across digital lending, identity and KYC, payments and enterprise integrations. Program discipline and product judgement in one person: I run programs end to end, from PMO to regulatory audits, and I own products from discovery to launch and growth. Agents and automations on Jira, ClickUp and Notion carry the routine, so teams spend their time on decisions. I also build AI products myself, from the user's problem to a measured, governed release: machine learning, RAG and agents, with guardrails a regulator would accept.

Portrait of Prashant Bharam, smiling with arms crossed, in a light blue shirt
Göteborg / GothenburgPMP certified
15M+

identity verifications shipped across four countries (Video KYC, bank verification, face match and liveness)

~60%

of loan-disbursal revenue powered by the integrations my team delivered

40K

active merchants on a credit line launched from zero in four months

15+

programs and products delivered, from 0→1 launches to regulatory audits

Program and product

Two disciplines, one operator.

I have run regulatory programs and shipped products at startups and a payments bank. I move between the two as the work needs, which is what an early-stage team usually needs.

Program management

Delivery that holds up under a regulator, a CXO or a Big 4 client.

  • PMO and Agile governance from zero: Scrum, Kanban, SAFe-style planning, RAID and OKRs.
  • Regulatory and compliance programs: RBI audit, ISO 27001/27701, SOC 2, VAPT, GDPR and DPDP.
  • Enterprise and government implementations, integrations and HRMS connectivity.
  • C-level steering committees, budgets, vendor spend and cost decisions.
1stRBI audit led, passed
-60%unplanned work
0critical incidents post-launch
Senior Program ManagerTechnical Program ManagerScrum MasterPMO Lead
Product management

Roadmaps tied to revenue, risk and what customers actually do.

  • Vision, roadmaps, PRDs and success metrics from discovery through launch and growth.
  • Customer discovery workshops, enterprise pilots and go-to-market demos.
  • Funnel analysis with SQL and KPI dashboards, and ROI-based prioritisation.
  • API and platform products: identity verification, B2B integration suite, employer platform, CRM and lending journeys.
40Kmerchants in 4 months
15M+verifications shipped
+18%retention, +15% ARPU
Senior Product ManagerProduct ManagerTechnical Product ManagerProduct OwnerAI Product Manager
AI-driven delivery

The program manager's day, rebuilt around automation.

Most of a program manager's week is routine: triage, chasing, reporting, risk logs. I build agents and automation rules for that layer in Jira, ClickUp and Notion, and keep the judgement for myself.

I do not only prompt AI. I build agents and automations and run them in daily delivery.

Automate the second time

The first time I do something, I learn it. The second time, I build the rule, template or agent that does it.

Humans keep the decisions

Agents draft, flag and remind. I own scope, risk calls, escalation and every number that reaches a CXO.

Evidence by default

Status, risks and decisions live in the tool, so audits, steering committees and handovers pull from one source.

Stage of the programWhat runs automatically or with an agentWhat I decideWhere
Intake and triageRequests arrive through a form. Automation creates the ticket, sets program, priority and owner, and an AI step drafts the acceptance criteria.Confirm scope and value before anything enters a sprint.Jira · ClickUp · LLM
Backlog and sprint planningAn agent checks the backlog before planning for missing estimates, duplicates, unclear criteria and blocked dependencies.Set the sprint goal and the trade-offs.Jira · ClickUp · LLM
Daily flowStatus rules move tickets, nudge owners on stale work and post a morning digest of what changed and what is blocked.Clear blockers and re-sequence work.Jira · ClickUp
Risk and RAIDAn agent scans ticket activity for risk signals and proposes RAID entries with the source linked.Own mitigation, escalation and the final rating.Notion · LLM
Status reportingThe weekly report is drafted from live ticket data and Notion pages, with RAG status and trend.Add judgement, check every number and send it.Notion · Jira · LLM
Steering committeesAI drafts the pre-read and the decision log from KPI dashboards and meeting notes.Frame the decisions and the asks for the C-level.Notion · LLM
Release and complianceRelease checklists and audit evidence for ISO 27001, SOC 2 and RBI are tracked as backlog items with automated reminders.Sign off go-live and audit readiness.Jira · Notion

Every agent has an owner, a defined scope and a trail in the tool. In regulated work that matters as much as the time saved.

AI projects · built and measured

I build the AI, then I measure it.

Three working products for regulated delivery and compliance. Each one starts like a product, with a real user and a problem I have lived, and is delivered like a program: tested on data it was not tuned on, a human in charge, and its weaknesses stated.

Flagship · Live demo · 2026

Program Risk & Compliance Copilot

The problem it solves

Program managers in regulated fintech lose hours every week to three jobs: spotting late work only once a release is already at risk, answering questions like “our cloud provider was down 6 hours, do we have to report it?” from hundreds of pages of law, and building the weekly steering-committee report by hand.

What it does

  • Flags delivery risk: ranks open tickets by how likely they are to miss their release, with the reasons for each.
  • Answers compliance questions: from the official DORA, GDPR and PSD2 texts, with the exact article, or says clearly when the law does not cover it.
  • Drafts the weekly report: status, top risks, regulatory checks with citations and decisions needed, ready for a human to approve.

How it works

  • Machine learning trained on 9,000 public Apache Jira tickets. The simple, explainable model beat a more complex one.
  • RAG over the EU texts: it first shows how it understood the question (typos, Swedish and real situations welcome), then answers only from the articles it found.
  • A Claude agent picks its own tools to investigate the portfolio and write the report.

Built for a regulated firm

  • Personal data hidden before any AI call; hidden instructions in tickets ignored and reported.
  • Every citation checked by code; nothing sent without a named approver; tamper-evident audit log.
  • Run as a full program for a fictional Swedish payment institution: business case, requirements, DPIA and DORA vendor assessment of the tool itself, Jira delivery, go-live plan.
100%correct article cited on held-out questions
88%of messy questions handled (typos, Swedish)
1.6×better than random at finding late work
$0.02per answer · $0.14 per weekly report

Open weakness, tracked as a go-live gate rather than hidden: answers sometimes add detail beyond the cited text. Free hosting, so the first visit after a quiet period takes about 50 seconds.

PythonClaude APIRAGscikit-learnAI agentsLLM evaluationGuardrailsJira
Open source · 2026

PM AI Toolkit: approval-gated agents for the PMO

The problem it solves

Weekly status reports and chasing meeting actions eat a program manager's week. AI can draft them, but an unchecked AI status sent to executives is a governance risk.

What it does

  • Status agent: turns a Jira export into a weekly red-amber-green status report, then sends it by email or Slack only after a named person approves.
  • Meeting-to-action: pulls owners, actions and due dates out of meeting notes.

How it works

Code counts, the model judges, a checker rejects answers not grounded in the data, and a person approves. Every step lands in a tamper-evident audit log. Works offline with a rule-based baseline; read-only Jira connector.

87.5%status accuracy on 24 labelled weeks
100%of answers grounded in the data
95automated tests in CI

Two modules built, more planned. Evaluated on synthetic data, with the limits published next to the results.

Live · work in progress · 2026

RBI KYC assistant: cited answers from the regulator

The problem it solves

India's KYC rules for lenders and banks are amended often. Compliance and product teams lose time finding the current paragraph, and risk acting on an outdated circular.

What it does

  • Answers KYC questions for NBFCs and commercial banks from the official RBI text, with the source paragraph and date.
  • Treats the newest circular as controlling, and can include the repealed 2016 text for history.
  • Says so when nothing relevant is found, instead of guessing.

How it works

Retrieval over 22 official RBI documents (the 2025 KYC Directions, their amendments and the 2016 Master Direction history), with every content update going through human review. A compliance aid, not legal advice.

22official RBI documents tracked
Citedevery answer shows its source
Refuseswhen the text does not cover it
How I run a program · sample artifacts

The documents behind the copilot, written as a real program would.

The copilot was delivered for a fictional Swedish payment institution, with the same artifacts I produce in regulated delivery. Open any of them.

How I take over a program or a product

The first 90 days, in three moves.

Whether it is a program in flight or a product that needs direction, I follow the same sequence: understand users and constraints before changing anything, stabilise before scaling, and leave it running without me.

Days 1–30 · See it clearly

Meet stakeholders, teams and customers. Rebuild the real picture from the tools and the data, not the slides: users and their problems, scope, dependencies, risks, budget and regulatory deadlines. Agree what “done”, “on track” and “success” mean, and publish one source of truth.

Days 31–60 · Stabilise and focus

Fix the cadence: planning, RAID, change control and a steering committee that takes decisions. Turn discovery into a prioritised roadmap with clear success metrics. Close the riskiest gaps first, especially anything a regulator or auditor will ask about, and automate reporting so status is pulled, not chased.

Days 61–90 · Ship and scale

Ship the first improvements, tie the roadmap and the program to outcomes in the business case, and coach leads and product owners to run the rhythm themselves. The goal is a program and a product that keep delivering when I step back.

Blog

What I am learning about delivery and AI.

Notes and learnings from building and running the projects above. All posts · RSS

Blog · 2026-11-16

Running an AI pilot like a regulated program

A tool that helps with compliance has to be compliant itself. How I took an AI assistant from idea to go-live gate with the governance a bank would expect.

Selected work

Eight programs and products, with the numbers.

Each one had a regulator, a deadline or a revenue line attached. Context, what I owned, and what changed.

programRegulatory · Refyne · 2025

Leading a first-ever central bank audit

Refyne had never been audited by the RBI. No playbook, five departments, one deadline.

  • Took sole ownership of a three-month readiness program across Finance, Technology, Operations, Security and InfoSec.
  • Mapped gaps, assigned owners, tracked them in a single risk log and reported weekly to leadership.
100%of identified gaps closed ahead of schedule
1staudit in company history, passed
programIntegrations · Refyne · 2023 to now

Enterprise integrations that carry the revenue

Large employers connect payroll and HR systems to Refyne's earned wage access product.

  • Led a three-person team delivering integrations for the Rajasthan Government, Deloitte, EY and Flipkart.
  • Supported HRMS patterns across API, hybrid and manual models: Darwinbox, PeopleStrong, SAP SuccessFactors, ADP, Workday.
~60%of loan-disbursal revenue runs on these
0critical production incidents after launch
programOperating model · Refyne and MyShubhLife

PMO and Agile, built from zero. Twice.

Two fast-growing startups where delivery process had to be built from scratch, under fast-moving priorities.

  • Introduced Scrum and Kanban, SAFe-style planning, RAID management and milestone reporting.
  • Embedded compliance work (ISO 27001/27701, SOC 2, VAPT, DPDP) into the backlog instead of running it on the side.
-60%unplanned work
+30%on-time delivery
~80%on-time at MyShubhLife
productDigital identity · IDcentral · 2022 to 2023

Video KYC that regulators accept

Banks, fintechs and telecoms in India, Nigeria, Vietnam and the Philippines needed remote customer onboarding.

  • Owned the roadmap for identity APIs: Video KYC, bank account verification (penny-drop and UPI checks), face match and liveness.
  • Ran customer discovery workshops and enterprise pilots, then turned feedback into prioritised requirements and sales demos.
15M+verifications delivered
-40%manual KYC effort
-30%time to integrate
product0→1 launch · Fino Payments Bank · 2022

A merchant credit line in four months

Fino wanted to lend to small merchants through two NBFC partners.

  • Owned the roadmap and launch of the credit line, then used funnel data to rework onboarding.
  • Cut credit onboarding, including credit-bureau checks, from about 15 minutes to under 5.
40Kactive merchants
$1M+disbursed in 4 months
+18%retention, with +15% ARPU
programExecutive alignment · Refyne · 2024 to now

Steering committees that release budget

Two Big 4 clients needed one place where C-level sponsors, risk and delivery met.

  • Facilitated monthly C-level steering committees and kept decisions, risks and dependencies visible.
  • Unlocked mid-six-figure budget commitments and led the ISO 27001/27701, SOC 2, VAPT and DPDP workstreams behind them.
+12%loan disbursals
2Big 4 clients, monthly C-level cadence
productPlatform products · Refyne · 2023 to now

Owning the product suite behind earned wage access

Employers and Refyne's own teams needed integrations, a CRM and an employer platform that worked together.

  • Own vision, roadmap and success metrics for the B2B Integration Suite, the FinSuite employer platform and an in-house CRM.
  • Take FinSuite's salary-on-demand workflows from PRDs and UAT to launch, settlement and reconciliation, so employer billing is audit-ready.
  • Replaced HubSpot with the in-house CRM and own the vendor spend decisions.
3products owned end to end
~$37Ksaved per year on CRM
productDigital lending · MyShubhLife · 2016 to 2022

Lending products, from QA lead to product owner

A digital lender with NBFC partners needed new credit products and a delivery rhythm that scaled.

  • Owned roadmaps and delivery for BNPL, revolving credit and term loans, leading 10 to 12 person squads.
  • Used funnel and drop-off analysis to prioritise journey changes, and built 8 to 10 financial-wellness partnerships.
$10M+disbursed, NPA held at 1.9%
35 to 40%first-loan activation
1M+users reached
Translated for Sweden

Built in India. Directly relevant to Nordic fintech.

Most of my work sits under Indian rules and products. The problems underneath are the ones Nordic fintechs and banks face too.

På svenska

Senior programledare och teknisk programledare med nio års erfarenhet av reglerad fintech: utlåning, KYC, betalningar och integrationer. Jag leder program hela vägen, från PMO till myndighetsgranskning, och bygger AI-verktyg med tydliga kontroller och mänskligt godkännande. Jag känner till DORA, GDPR och PSD2 och arbetar redan med dem i mina projekt. Jag bor i Göteborg, har fullt arbetstillstånd och lär mig svenska.

What I worked withClosest Nordic parallel
RBI auditA supervisory review. The RBI is both central bank and financial regulator, so think Riksbank and Finansinspektionen combined.
Account AggregatorIndia's RBI-regulated open banking framework for consent-based data sharing, comparable to PSD2 account access.
UPIReal-time account-to-account payments, comparable to Swish.
Video KYCRemote customer onboarding under AML rules, with identity checks like face match and liveness.
NBFCA licensed non-bank lender, similar to a Swedish finance company (finansbolag).
DPDP ActIndia's data protection law. I have worked to both DPDP and GDPR requirements.
Earned wage accessEmployees draw part of earned salary before payday, settled through the employer's payroll.
Experience

From QA lead to program owner.

Nine years in fintech delivery, starting in test engineering. That is why I read a release plan the way an engineer does.

Feb 2025 to present
Sep 2023 to Feb 2025
India

Senior Program Manager / Product Owner

Refyne · Earned wage access and financial wellness · PMO and B2B programs

  • Promoted after introducing Agile ways of working from scratch and lifting delivery predictability by about 25%.
  • Own product vision, roadmaps and success metrics for the B2B Integration Suite, the in-house CRM and the FinSuite employer platform, from PRDs and UAT to launch, settlement and reconciliation.
  • Built agents and workflow automations on Jira, ClickUp and Notion that run triage, reporting and risk tracking day to day.
  • Govern organisational OKRs and run KPI dashboards for the leadership team.
Sep 2022 to Sep 2023
India

Technical Product Manager, Digital Identity and KYC

IDcentral · Digital identity and risk

  • Owned APIs for Video KYC, bank verification and face match across BFSI, fintech and telecom clients in four countries.
  • Worked with sales, engineering and compliance on go-to-market demos that shortened enterprise sales cycles.
May 2022 to Sep 2022
India

Product Manager, Lending and Wealth

Fino Payments Bank

  • Launched a merchant credit line with two NBFC partners, plus roadmaps for gold savings and insurance.
May 2019 to May 2022
Nov 2016 to May 2019
India

Product Manager, promoted from QA Lead

MyShubhLife · Digital lending platform

  • Led 10 to 12 person squads shipping BNPL, revolving credit and term loans. $10M to $15M disbursed with NPA held at 1.9%.
  • Built 8 to 10 financial-wellness partnerships reaching 1M+ users, with 35 to 40% first-loan activation.
  • As QA Lead, moved the organisation to test-driven development with automated regression and API testing.
How I work

Toolbox and working style.

Delivery

ScrumKanbanSAFePRINCE2 AgilePMO setupRAIDRelease governanceOKRs

Fintech and compliance

Digital lendingBNPLEarned wage accessKYC / AMLOpen bankingISO 27001/27701SOC 2GDPR / DPDP

Technical

REST APIsSSO / SAMLSFTPSQLCI/CDPostmanJiraConfluenceClickUpNotion

AI and automation

AI agentsWorkflow automationJira automationClickUp automationNotionPrompt engineeringGenerative AIAgentic coding toolsRAGLLM evaluationClaude APIscikit-learnAI guardrails

This site was built with an agentic coding tool in one sitting.

Credentials

Certified, educated, international.

Certifications

  • Project Management Professional (PMP), PMI
  • Certified Scrum Product Owner (CSPO)
  • SAFe 6.0 Agilist
  • PRINCE2 Agile Practitioner
  • Generative AI for Project Managers (Coursera)

Education

  • MBA, Liverpool John Moores University, UK
  • PG Program in Management, IMT Ghaziabad
  • BBA (Computer Applications), Sinhgad Institute of Management, Pune
  • IB Diploma, UWC Adriatic, Italy (Davis UWC Scholar)

Languages

  • English, professional
  • Swedish, basic and improving
  • Italian, basic
Conversations I enjoy

Let's talk over coffee.

I would rather have a good conversation than send a document. Coffee in Göteborg or a 30-minute call, on any of these.

Agentic finance

Where AI agents belong in financial services

Which decisions agents can carry today, which stay with people, and how to keep humans in control of the numbers that reach a board.

Regulated AI

Shipping AI that a regulator would accept

DORA, GDPR and the EU AI Act in practice: evaluations, audit trails and guardrails that are built in from day one, not bolted on before an audit.

Fintech delivery

Startup speed with banking discipline

Lending, KYC, payments and integrations: how to move fast without surprises in production, at audit time or at the steering committee.

Prashant Bharam

Let's grab a coffee

Building something in fintech or agentic finance, or running a program you want to compare notes on? I am always up for a coffee in Göteborg or a call. Based in Göteborg, with full work rights in Sweden.