Real-time & communications
WhatsApp, voice, video, SMS and PSTN/VoIP at carrier scale. Session handling, failover, media pipelines — and the observability to prove it holds under load.
Available for consulting · Q4 2026
I build systems that don't break when things get real.
Product Engineer at Acefone. Six years shipping real-time communications, distributed backends and AI products — for teams in India, Canada, Belgium and France, where one second of downtime costs real money.
Usually replies within a day · hello@speedsharma.com
6+
years in production
100K+
users scaled to
4
countries shipped for
9
open-source packages
2
published books
what i do
Hire me for the parts that are expensive to get wrong.
AI product engineering
LLM gateways, RAG pipelines, agents and tool-calling — with cost tracking, guardrails and evals, so it survives contact with real users.
Backend architecture at scale
Microservices that stay debuggable. Event-driven design, data modelling, queueing — and a realistic migration path out of the monolith you already have.
Observability & developer experience
Structured logging, OpenTelemetry tracing, canonical log lines and internal tooling that makes your team faster than another hire would.
selected work
Systems that had to hold.
Asia's first WhatsApp IVR, inside one agent inbox
Acefone was moving off a legacy PHP monolith onto MERN microservices. I took the greenfield work and built Interaction Hub — one inbox where an agent handles voice, WhatsApp messaging and WhatsApp Calls without switching tools.
Zero to 100,000 users as a founding engineer
Founding team member at a stock advisory platform. I built the product from nothing, led three developers, and owned the real-time market data pipeline — then kept shipping unpaid for two months when regulation put the funding at risk.
A Canadian property platform, built solo
Six months, one engineer, a full platform for the Canadian market — role-based access control, real-time messaging between buyers and agents, and an architecture chosen so a single person could keep it running after handover.
WhatsApp automation, run by one person
WhatsApp and Instagram automation for businesses. Product, infrastructure, billing, deliverability and support — all mine. The constraint was never scale; it was that every system had to be boring enough for one person to operate at 2am.
open source
The tools I wished existed, so I built them.
flagship
Logixia
An async-first TypeScript logger built for observability 2.0 — wide events and canonical log lines instead of string soup, OTLP OpenTelemetry export, runtime-adjustable levels, PII redaction, adaptive sampling and multi-transport delivery.
wide events
otlp export
pii redaction
adaptive sampling
runtime log levels
multi-transport
- works with
- nest · express · fastify · bun
- license
- MIT
reixo
TypeScript
Type-safe HTTP client with built-in retry, queueing and resilience.
reins
TypeScript
Structured concurrency and cancellation — scoped tasks, timeouts, no DSL.
exostate
TypeScript
Fully type-safe state management for frontend and backend alike.
mongozen
TypeScript
Type-safe MongoDB aggregation pipeline builder for Nest, Express, Node.
react-scale
TypeScript
CLI that scaffolds production-ready React apps from a battle-tested template.
comprexia
C++
Compression library with JSON-aware optimisation and Node N-API bindings.
writing
The parts that don't fit in a tweet.
Idempotency in Distributed Systems — Making Any Operation Safe to Retry
Learn idempotency key design, idempotency stores with TTL, request fingerprinting, and CQRS deduplication patterns for safe retries.
Mar 2026
LLM Observability in Production — Tracing Every Token From Request to Response
Master end-to-end LLM observability with OpenTelemetry spans, trace correlation, adaptive sampling, and anomaly detection to catch production issues before users do.
Mar 2026
Logging Everything and Nothing Useful — The Noise Problem
Your logs are full. Gigabytes per hour. Health check pings, SQL query text, Redis GET/SET for every cached value. When a real error occurs, it's buried under 50,000 noise lines. You log everything and still can't find what you need in a production incident.
Mar 2026
what they say
The people who had to work with me.
“Sanjeev's architectural decisions improved our system performance dramatically.”
Acefone
“Sanjeev delivered an exceptional platform that transformed our business operations.”
CEO, E-mopro
“A 5-star rating held across every engagement, with clients in four countries.”
client rating, maintained
how it works
Three steps. No discovery theatre.
01
A 30-minute call
You describe the system and where it hurts. I ask the awkward questions. If I'm not the right person for it, I'll say so and point you to someone who is.
02
A scoped plan
Within a week: written architecture, the risks I'd worry about, sequencing and a fixed price. Yours to keep whether or not we work together.
03
Ship it
I embed with your team or run it solo. Weekly demos, real commits, and documentation your engineers can own long after I'm gone.
Got a system that has to hold?
Tell me what you're building and what worries you about it. I read every message, and I reply within a day.
Gurugram, India · IST · comfortable with US and EU overlap