An engineer who
refuses.
I'm Sagar — a full-stack engineer with an MSc in Robotics. I build production distributed systems and ML pipelines end-to-end, from Kafka-backed microservice platforms to ROS-based autonomous robots — and I've hardened national water infrastructure along the way.
I think out loud. I prototype before I argue. The interesting work rarely fits in last year's stack, so I never let myself settle into one.
Start a conversation →B · Working principles
Curious by default
If I haven't tried it yet, I want to. The unread paper, the unfamiliar runtime, the weekend reading list — all on the same shelf as the day-job.
FoundationFreedom to build
I do my best work when given a problem, not a process. Hand me the question; I'll bring back the shape of the answer.
ApproachCross-pollinator
AI ↔ robotics ↔ control ↔ web. Ideas from one routinely repair the other. The disciplines are friends, even when their tooling argues.
MethodOutcome over ego
Ship the thing. Iterate. The first draft is a hypothesis, not a trophy. I'd rather be edited than admired.
AttitudeC · Trajectory
Founding Engineer — Clearbox · Remote / London
- Architected and shipped a 4-layer platform of 20+ Dockerized microservices: React / Next.js frontends, seven FastAPI services, a Node / Express RS256 JWT auth service, Kafka, a MinIO medallion data lake, and Postgres / MongoDB / Redis.
- Built the data pipeline end-to-end — browser capture → Kafka streaming → ETL scoring → Postgres analytics, with a raw → bronze → silver → gold lake feeding role-based admin dashboards.
- Fronted all 20+ services behind a single Nginx reverse-proxy gateway; ran parallel dev / prod Docker Compose stacks shipped by a GitHub Actions CI/CD pipeline with drift-proof, idempotent SSH deploys.
- Stood up a Prometheus + Grafana + node-exporter observability stack with auto-provisioned dashboards.
- Selected top 5% in Bank of America's incubation programme.
SCADA Engineer — Thames Water · Reading
- Built Python / SQL automation pipelines across 300+ water and wastewater sites — cutting reporting time 50% and eliminating QC errors.
- Delivered end-to-end SCADA integration for critical national infrastructure: RTU configuration and telemetry commissioning over TCP/IP and DNP3-WITS.
- Led technical project delivery — document pipelines, compliance against asset standards, and zero-error site handovers.
ML / Data Engineer — University of Birmingham
- Built an end-to-end ML pipeline on 500 robotic-surgery records — PCA to isolate principal clinical-outcome features, then classical models for predictive insight.
- Found two clinically meaningful clusters in da Vinci surgical data (validated at silhouette 0.77 / 0.98); regression models reached R² 0.91 / 0.77 for procedure duration and arm active time.
- Led data collection with senior surgeons and drove KPI definition aligning clinical and engineering priorities.
MSc Robotics (Merit) — University of Birmingham
- Dissertation on ad-hoc multi-agent teamwork — neuro-symbolic reasoning (SPARC ASP + PyTorch behaviour cloning) with BREMEN offline RL.
- Achieved a 30% inference improvement over baseline, coordinating with unknown teammates at game speed and no shared policy.
Software Engineer — Cognizant · Bangalore
- Built REST APIs and SQL schemas for microservice prototypes in Java / Spring Boot; optimised queries for 20% faster performance.
BTech Mechanical Engineering — Kerala Technical University
- Graduated 7.93 / 10, with the engineering grounding that still shapes how I think about systems.
D · Stack — in rough order of affection
Pitch me the wildest version.
We can scope down from there. I'd rather hear the ambitious version once and trim it together than guess at what you're really after.