I'm a backend engineer. Most of my 11+ years has been on systems where losing or double-handling a message means real money goes the wrong way: payments, banking, and a data platform running 30M+ rows in Postgres. I tend to care most about the boring parts, like catching a bad input and retrying it instead of letting it quietly break something at 2am.
I also build agentic AI systems, and I build them the way I build backend systems: retries, idempotency, structured state, evals. ChatFormula1 (LangGraph + Pinecone RAG, live at chatformula1.com) and Assay (FinResearch AI — multi-agent equity research on LangGraph/FastAPI/pgvector, live on Hugging Face Spaces) are both deployed with CI, tests, and rate limiting. Both are below.
Right now I'm a Software Engineer at Agilent on the manufacturing side, where part of the job is helping the team actually use AI day to day. Before that: HG Insights, Capital One (through Cognizant), and AurionPro.
prateekmulye.dev · LinkedIn · Email
ChatFormula1 — a single-agent agentic RAG assistant for Formula 1 (live)
One agent routes between a vector store and live web search, then answers over what it retrieves. Built with production hygiene: auth, rate-limiting, tests, structured logging, caching, graceful degradation, CI/CD, and Pydantic-typed state.
LangGraph · Pinecone · Tavily · FastAPI · Pydantic
Assay (FinResearch AI) — a multi-agent equity-research system (live on Hugging Face Spaces)
A LangGraph manager fans out to specialized research agents in parallel, then an analyst and a reporter write a verdict over what was retrieved. Every step emits structured JSON so the agents stay grounded instead of guessing. Grew out of my submission to the SuperDataScience CP044 community project.
multi-agent · LangGraph · FastAPI · pgvector · RAG · Python
slipstream-f1-strategist — an event-driven race-strategy service
A reactive API publishes a request to Kafka, a consumer runs a deterministic simulation, and the result is written back exactly once. The NDA-free analogue of my Capital One reliability work: bounded retry with backoff and idempotent processing, so a bad message is retried sensibly and never processed twice.
Java 21 · Spring WebFlux · Kafka · PostgreSQL · OpenTelemetry · Testcontainers
| Distributed systems | Apache Kafka, PostgreSQL at 30M+ rows, idempotent recovery, DLQ + bounded retry, service contracts |
| Backend | Java (Spring Boot), Elixir (Phoenix), Python, GraphQL and REST |
| Applied AI | LangGraph, RAG, Pinecone, Pydantic-typed state, multi-agent orchestration |
| Platform | Docker, Kubernetes, Terraform, OpenTelemetry, Datadog, AWS |
More at prateekmulye.dev.