Seven years of shipping to production
Shrey
Alkhaniya
I build web systems that hold up under load.
Full-stack engineer working in TypeScript, Next.js and Node — and I can show you the numbers behind every claim on this page.
Who I am
Most of my work has been on systems where slowness and errors cost somebody real money — a payments bank, a telecom portal, a benefits platform. That has shaped how I build: measure first, ship the smallest thing that moves the number, then prove it moved.
01
Measure, then move
I don't ship performance work on a hunch. Correlation IDs, OpenTelemetry spans and structured logs come first, because the fix is usually obvious once the trace is readable. The 504s I chased across a React → Node → Django chain took weeks to find and an afternoon to fix.
02
Fast where it's hard
Most of my users have been on constrained networks and older hardware — branch bank staff, not engineers on fibre. That makes payload size, compression and interactivity real constraints rather than lighthouse-score vanity, and it's why the UIs I build stay responsive and accessible by default.
03
Comfortable at the seams
Message brokers, gRPC contracts, serverless functions, vector indexes — the parts where one system hands work to another. I like this layer because it's where architecture decisions become visible, and where a wrong one gets expensive quietly.
What that produced
Four numbers I can walk you through line by line, with the commit history to match.
- Messages/day
- ~10k
- through the Message Center
- Daily users
- ~6k
- across three portals
- Faster P95
- 40%
- 7.1s → 4.3s on low bandwidth
- Fewer errors
- 60×
- 1.2% → 0.02% error rate
Career trace
Seven years, read as a waterfall.
Bar widths are proportional to actual tenure. Each nested span is one shipped outcome and the number it moved.
Trajector
Senior Software Engineer
Aug 2024 — Present · 2 yr 1 mo
Message Center
~10k msgs/day · ~6k daily users · ~1,000 support hours saved/month
Customer tooling
+29% page interactivity
Jio
Software Engineer II
Feb 2023 — Jul 2024 · 1 yr 6 mo
Design-token pipeline
~2 hrs/release → near-zero
Scheduled digest service
~95% cost reduction
504 investigation
error rate 1.2% → 0.02%
Paytm Payments Bank
Software Engineer
Jun 2019 — Jan 2023 · 3 yr 8 mo
Low-bandwidth performance
7.1s → 4.3s (40% faster)
RBI-compliant testing platform
~1,500 employees onboarded in week one
Data-loss fix
~10 incidents/month → 0
Stack
What I reach for.
Ordered by how central each one is to how I actually work, not by how it looks on a keyword scan.
Frontend
- TypeScript
- React
- Next.js
- App Router / RSC
- Server Actions
- Tailwind CSS
- Web Vitals
- Accessible UI
Backend
- Node.js
- Express
- REST APIs
- gRPC / Protobuf
- RabbitMQ
- AWS Lambda
- Django
- Event-driven services
Data
- PostgreSQL
- pgvector
- MongoDB
- Mongoose
- Drizzle ORM
- Schema & index design
Platform
- Docker
- Azure Functions
- GitHub Actions
- OpenTelemetry
- Structured logging
- Correlation IDs
AI / LLM
- RAG pipelines
- Embeddings
- Top-K vector retrieval
- HNSW
- Prompt grounding
- Ollama
Featured work
Two things I'm building right now.
Both are in progress and neither is public yet — so these cards describe only what runs on my machine today. Links go up the day they ship.
- 01In progress
AskMyDocs
Retrieval-augmented Q&A over your own documents.
A RAG system built around the part that actually decides answer quality: the retrieval layer. Per-user chunks with 1536-dimension embeddings sit behind an HNSW cosine index, with user_id denormalised onto chunks so a tenant-filtered vector search stays a single index scan rather than a join.
TypeScriptNext.jsPostgreSQLpgvectorDrizzleDockerOllama - 02In progress
ChatQ
The same chat service, modelled two ways: broker and RPC.
A multi-client pub/sub chat service on a RabbitMQ fanout exchange with per-client exclusive queues — then the same contract re-modelled in gRPC, specifically to feel the difference between broker-based delivery and direct RPC rather than read about it.
Node.jsRabbitMQgRPCProtobufpino
Open to work
Hiring for something that has to stay fast under load?
I'm open to senior and staff full-stack roles, remote or in India. The quickest way to start is a short email about the problem you're trying to solve.