Forward
Deployed
Engineer.
0%
Forward Deployed Engineer · San Diego, CA
Rakesh Suryavanshi Engineer.

I partner with engineering teams to turn complex technical problems into shipped solutions — from API integrations to production-grade AI deployments.

I can |
30+
Projects
8+
Years
MS
Computer Sci
Scroll to explore
FastAPI
RAG Pipelines
pgvector
Docker & Kubernetes
LLM Integration
REST & GraphQL
Next.js 14
AWS · GCP
Celery · Redis
Spring MVC
PostgreSQL
CI/CD Pipelines
FastAPI
RAG Pipelines
pgvector
Docker & Kubernetes
LLM Integration
REST & GraphQL
Next.js 14
AWS · GCP
Celery · Redis
Spring MVC
PostgreSQL
CI/CD Pipelines
RS
Rakesh Suryavanshi
Forward Deployed Engineer
APIs AI/LLM DevOps Full-Stack Open Source
30+
Projects Shipped
8+
Years Exp.
MS
Cal State LA
Problems Solved
About Me

Building at the edge of engineering & delivery

I'm Rakesh Suryavanshi, a forward deployed engineer with an MS in Computer Science from Cal State LA. I've spent 8+ years building production software across the full stack.

I partner with engineering teams to turn complex technical problems into shipped solutions. My sweet spot is the gap between a product vision and working software — I embed fast, understand the real problem, and ship.

Currently focused on AI/LLM deployments, API integrations, and DevOps for startups that need someone who can go from zero to production without hand-holding.

Selected Work

Projects & Engagements

★ Open Source · Featured

Multi-Tenant RAG Engine

Production-grade Retrieval-Augmented Generation service with strict multi-tenant data isolation. Async-first — FastAPI, Celery, SSE streaming, Redis caching, pgvector HNSW. Zero API costs via local Ollama.

FastAPIpgvector CeleryRedis Ollama / qwen2.5:14b Next.js 14Docker PrometheusJWT + RBAC MinIOAlembic
View on GitHub ↗
Architecture
Next.js 14 → FastAPI → PostgreSQL + pgvector → Redis → MinIO → Celery → Ollama qwen2.5:14b
Multi-tenant — every query scoped by tenant_id, BOLA structurally impossible
Async ingestion: upload → Celery → chunk → embed → pgvector HNSW index
SSE streaming, Redis query/embedding cache, sliding-window rate limiting
Prometheus metrics, structured logging, health endpoints — production-ready
Earlier Projects
01
MyDentist Portal

Full-stack healthcare portal — live appointments, discharge summaries, lab results, email notifications, jQuery chat, Spring Security auth.

Spring MVCHibernateAJAX/jQueryJSPSpring SecurityMySQL
2017 GitHub
02
Beautiful Data — Social Analytics

Data acquisition from Twitter & Facebook APIs. MongoDB storage, Amazon Elasticsearch for indexing, real-time social analytics visualisation.

Amazon ESTwitter APIFacebook APIMongoDBNode.js
2016 GitHub
03
CS Crawlers — Web Search Engine

Concurrent web crawler with HTML parsing, JSON metadata extraction, full-text indexing and search via Apache Lucene with multithreaded request handling.

Apache LuceneJavaJSONMultithreading
2016 GitHub
04
Watermark Image Editor

Custom ASP.NET image editor with watermark overlay, colour/font/size controls, Java Servlet backend and JSP/JSTL frontend with Bootstrap UI.

ASP.NETJava ServletJSP/JSTLBootstrap
2016 GitHub
Technical Stack

Everything I work with

From enterprise Java systems to modern AI/LLM pipelines and cloud infrastructure — technologies I've used in production.

AI / LLM
RAG PipelinesFastAPILangChainpgvectorOllamaOpenAIPrompt Eng.SSEEvals
Backend
PythonJavaSpring MVCNode.jsREST APIsGraphQLCeleryPostgreSQLMongoDBRedis
Infrastructure
DockerKubernetesTerraformAWSGCPGitHub ActionsMinIOPrometheusNginx
Frontend
Next.js 14ReactTypeScriptTailwindjQueryBootstrapJSP/JSTLHTML/CSS
How I Work

The FDE playbook

01 — Discover
Read before the call

I review the codebase before the first meeting. The stated problem is rarely the real one.

02 — Scope
Define what done means

Crisp spec, clear success criteria, explicit boundaries. No surprises mid-engagement.

03 — Ship
Working software fast

Bias toward shipping and iterating. Observability and rollback plans from day one.

04 — Handoff
No black boxes

Docs your team reads. Code your team maintains. Done when the next engineer owns it.

Get In Touch

Let's buildsomething great.

Hard integration, AI product to productionise, or infrastructure that needs to actually work — let's talk. Available within 1–2 weeks.