The Role
Build and operate production AI/ML and agentic solutions for banking and fintech partners. Diagnose distributed-system issues, develop tested fixes, automate recurring incidents, and own deployments through monitoring and continuous improvement. Collaborate directly with clients, engineers, product teams, and regulators while managing high-pressure escalations. Implement LLM, RAG, tool-calling, evaluation, observability, security guardrails, auditability, and human approval workflows for regulated financial systems.
Summary Generated by Built In
The Impact You’ll Drive
As a Forward Deployed Engineer (FDE), you’ll work at the intersection of AI/ML, production engineering, product and our banking and fintech partners. You’ll stay close to live systems, using code and AI to solve real fintech problems and keep mission-critical credit and payment programs running reliably at scale.
You’ll work directly with partners, program teams and regulators to learn how they operate, understand their workflows, find the problems that matter most and define clear success metrics.
Then you’ll deliver a solution quickly. That might be a scoped code change, an automation, a workaround, or an agentic system that stops recurring the same problem. You’ll take ambiguous, high-stakes problems from first diagnosis to a production system the partner depends on, and you’ll
own it after it goes live.
Success in this role requires strong engineering skills, a knack for solving operational problems, clear communication and the ability to turn ambiguity into working code quickly. You won’t just recommend solutions. You’ll build them and run them in production.
The Hats You Will Wear
- Lead discovery with bank partners, engineers, and product managers to understand operational workflows and deployment requirements, identify the real bottleneck, and define clear, measurable success metrics.
- Own technical problem-solving for live credit card, UPI, CLOU, lending tech, BRE, forex, rewards, ledgering and reconciliation programs across multiple bank partners, often juggling several escalations in flight at once.
- Build agentic debugging solutions for complex production issues across services, APIs, and data layers, and drive each issue to root cause and a permanent fix within strict SLAs.
- Design, build, and deploy AI/ML and agentic solutions in production. This includes LLM + RAG over runbooks, logs, and tickets, plus tool-calling agents (LangChain, LangGraph,CrewAI, MCP, ACP) that triage escalations, point to likely root causes, draft RCAs and
- automate recurring fixes. The goal is self healing systems.
- Ship fast, then harden. Deliver a working first version in days, then make it reliable over weeks with proper rollout plans, monitoring, and iteration based on real usage.
- Own production code changes end to end: read an unfamiliar service, make a scoped, tested fix, and ship it through a bank-grade change process: code review, audited commits, synthetic or masked data, rollback plan.
- Build AI guardrails for a regulated environment: human-in-the-loop approval for anything touching money or customer data, least-privilege access, prompt-injection and data-leakage controls, and full audit trail.
- Own the full lifecycle of the agents and harnesses you build: evaluation (groundedness, retrieval quality, tool-call accuracy, regression), controlled deployment, and observability (quality, latency, cost, drift, errors, confidence scores).
- Lead post-incident RCAs that produce clear, actionable insights. Tell a true product gap, which routes to developers apart from an implementation gap, which you own and close.
- Work directly with B2B clients (banks and fintechs) to automate near-real-time outage notifications and build trust under pressure.
- Feed what you learn in the field back into the platform and the core product, turning one-off fixes into reusable patterns so each deployment makes the next one easier.
- Operate effectively in dynamic environments with evolving business and deployment requirements.
The Perfect Fit
- 2–3 years of hands-on software engineering experience in B2B SaaS, fintech or payments, with real production code and AI/ML work to your name.
- B.E./B.Tech in Computer Science or AI & ML, with strong coding, data structures and algorithms fundamentals.
- Proficiency in at least one of Python, Typescript, Java or Go; able to find your way around an unfamiliar codebase, make a scoped, tested change and ship it using GitOps and LLMOps practices.
- Hands-on experience building AI/ML applications: LLM APIs, context engineering, embeddings, RAG, structured outputs and tool calling. You’ve built or operated at least four real agentic workflows in high-stakes production (orchestration, memory, plan–act–observe loops).
- Understanding of LLM evaluation, observability and guardrails, and why they must pass before a production release.
- Strong hands-on SQL skills, including joins, aggregations and tracking down data mismatches, plus a solid understanding of how requests and responses work across REST/HTTP, JSON-RPC and GraphQL APIs.
- Experience with monitoring and observability tools (Grafana, Prometheus, Kibana, ELK/EFK, OpenSearch) and a proven ability to debug distributed-systems issues.
- Understanding of system design basics (microservices, queues, retries, idempotency, failure handling) and working knowledge of databases (PostgreSQL preferred).
- Comfortable working with at least one cloud platform (AWS, GCP or Azure) and Kubernetes.
- Proven ability to handle several high-pressure bank escalations at once, calmly and through to resolution, with strong accountability, ownership and clear stakeholder communication.
Your Edge Over the Rest
- Experience with payments, banking or fintech systems: UPI, card networks (Visa, Mastercard, RuPay), Credit Line Over UPI, Credit Card Management systems, ledger systems and reconciliation flows.
- Familiarity with NPCI’s UPI ecosystem or RBI’s digital lending guidelines, and a strong instinct for handling regulated data.
- Depth in agentic and GenAI tooling (MCP servers, LangGraph, CrewAI, embeddings, FastAPI), with an eye on monitoring and inference costs.
- You don’t just fix issues. You prevent them, and you automate recurring ones out of existence.
- AI-native: you actively use AI tools in your own work and push for an AI-first approach in production by default.
- Full ownership: a problem is yours until it’s solved in production, not just until the code is merged.
- Highest standards: shipping 80% isn’t enough for you. You care about correctness, reliability, guardrails and edge cases, because these are a bank’s production systems.
The Problem We’re Solving
Financial institutions today are held back by legacy systems that are slow, rigid, and expensive to scale. Launching or evolving credit, lending, and UPI products often takes months, requires heavy engineering effort, and limits the ability to create personalized customer experiences.
At the same time, customer expectations have changed - speed, flexibility, and tailored financial products are no longer optional. Banks and fintechs need infrastructure that allows them to innovate quickly, adapt continuously, and scale without friction.
This is where we come in.
At Vegapay, we are building modern, configurable fintech infrastructure that enables banks, NBFCs, and enterprises to design, launch, and manage credit and payment programs with ease. Our platform brings together flexibility, speed, and control - helping our partners unlock new growth opportunities and deliver personalized banking experiences at scale.
The Opportunity Ahead
- Solve real-world fintech problems at scale with high ownership and impact
- Work directly with top banks and cutting-edge credit infrastructure
- Fast-paced, high-learning environment with strong growth opportunities
Skills Required
- 2-3 years of hands-on software engineering experience in B2B SaaS, fintech, or payments
- Bachelor of Engineering or Bachelor of Technology in Computer Science or Artificial Intelligence and Machine Learning
- Strong coding, data structures, and algorithms fundamentals
- Proficiency in at least one of Python, TypeScript, Java, or Go
- Experience finding, testing, and deploying scoped changes in unfamiliar codebases using GitOps and LLMOps practices
- Hands-on experience building AI/ML applications using LLM APIs, context engineering, embeddings, RAG, structured outputs, and tool calling
- Experience building or operating at least four real agentic workflows in high-stakes production environments
- Understanding of LLM evaluation, observability, and guardrails
- Strong SQL skills, including joins, aggregations, and investigating data mismatches
- Understanding of REST/HTTP, JSON-RPC, and GraphQL APIs
- Experience with monitoring and observability tools such as Grafana, Prometheus, Kibana, ELK/EFK, or OpenSearch
- Ability to debug distributed-systems issues
- Understanding of microservices, queues, retries, idempotency, and failure handling
- Working knowledge of databases, preferably PostgreSQL
- Experience with at least one cloud platform: AWS, GCP, or Azure
- Experience with Kubernetes
- Ability to manage multiple high-pressure bank escalations with strong ownership and stakeholder communication
- Experience with payments, banking, or fintech systems, including UPI, card networks, credit, ledgers, or reconciliation
- Familiarity with NPCI's UPI ecosystem or RBI digital lending guidelines
- Experience with MCP servers, LangGraph, CrewAI, embeddings, or FastAPI
Am I A Good Fit?
Get Personalized Job Insights.
Our AI-powered fit analysis compares your resume with a job listing so you know if your skills & experience align.
Success! Refresh the page to see how your skills align with this role.
The Company
What We Do
Vegapay is an Indian B2B fintech company providing software and tools for banks, NBFCs, and other financial institutions. It helps institutions launch flexible card and credit programs and modernize lending operations through a scalable, rapid-deployment platform. Its mission is to make personalized banking experiences and financial innovation more accessible regardless of institution size, while addressing the cost and implementation constraints of legacy systems.
