Skip to content
raj.bhoyar
Raj BhoyarRaj BhoyarOpen to backend roles

Four services.One engineer.

I'm Raj. I design, ship and get paged for every backend service behind a wellness platform used by ~9,000 students.

See the work Résumé
Scroll or tap to take it apart
✓ VERIFIED0 / 24
Missed crisis phrasings, down from 21 of 27AI safety guardrails →
✓ VERIFIED9,000
Active users in the pilotAI safety guardrails →
✓ VERIFIED150+
Automated deploys, ~4 min median push to liveLLM layer and capacity →
✓ VERIFIED10K+
Daily active usersGaming platform performance →
01 / Work

Case studies

Production systems, with the decisions and the numbers behind them.

01
AI safety guardrailsMissed crisis phrasings went from 21 of 27 to 0 of 24 test cases, with zero false alarms.Sole Backend Engineer · Hapminds · 2026 · NDA
0 / 24Missed crisis phrasings, down from 21 of 27
02
LLM layer and capacityhap_ai: one LLM interface with Gemini failover, a daily token cap, and model calls moved out of database transactions.Sole Backend Engineer · Hapminds · 2026 · NDA
200kTokens per user per day, enforced in Redis
03
AurNexa marketplace backendBackend design for a pre-launch B2B construction-materials marketplace: 9 bounded contexts, 5 in the MVP.Freelance Backend Engineer · AurNexa · Aug 2026 - now
9Bounded contexts, 5 in the MVP
04
Gaming platform performanceSlow PostgreSQL queries on high-traffic endpoints, fixed with query plans and new indexes.Full Stack Developer · Games World League · 2024 - 2025
10K+Daily active users
Side projects
RAG Evaluation HarnessPython harness that measures which embedding model and reranker retrieve best: accuracy and latency side by side.Python · RAG · Embeddings · RerankersMental-Wellness Assessment AgentChats with a user, picks the right screening test, scores it, and connects them to a psychologist.LLM agent · Python · GuardrailsAgentic assistantsMulti-agent assistants built with CrewAI.CrewAI · Python · Agents
02 / Experience & stack

Where I've shipped

From query tuning to owning a production GenAI backend.

Mar 2026 - Now
Software Development Engineer · HapmindsOnly backend engineer on 4 production services for ~9,000 active users. Safety guardrails cut missed crisis phrasings from 21/27 to 0/24; provider-agnostic LLM layer with failover and a per-user token cap; RAG with pgvector; 150+ automated deploys.FastAPI · Fastify · pgvector · Redis · RabbitMQ
Aug 2026 - Now
Freelance Backend Engineer · AurNexaDesigned a construction-materials marketplace backend as 9 bounded contexts. Saga with compensating steps over two-phase commit; transactional outbox on PostgreSQL instead of a separate broker.Architecture · PostgreSQL · Saga · Outbox
Jun 2025 - Feb 2026
Software Development Engineer · NNIITFastAPI services and LLM-powered learning features (RAG, multi-agent workflows) for ~1,000 active users; owned monitoring, alerting and incident fixes.FastAPI · RAG · Multi-agent · On-call
Stack
Languages
TypeScriptJavaScriptPythonSQLGo
Backend & data
FastifyExpressFastAPIPostgreSQLMongoDBRedisRabbitMQpgvector
AI
OpenAIGeminiRAGRerankingMulti-agentLLM evals
Infra
DockerGitHub ActionsDigitalOceanNginxLinuxBash
03 / About

I like making complex systems simple.

I'm a software engineer who works where backend systems meet the people using them: the queues, APIs, and AI pipelines that have to stay fast and correct under real load.

GenAI in production
Eval-gated GenAI service with two levels of safety checks at Hapminds.
Sole backend owner
Design, deploy and on-call for four production services.
Architecture
Sagas, outboxes and bounded contexts for a freelance marketplace backend.
Reading list · what stuck
04 / Contact

Let's talk.

Have a role or a project in mind? I'd like to hear about it.

LinkedIn ↗GitHub ↗Résumé ↗