Open to opportunities

Utkrisht Thakur

@utkrishtthakur

Message

Backend & ML Infrastructure engineer building fault-tolerant systems and production-grade AI pipelines.

India

Message

What I'm looking for

I’m looking for a role where I can build backend systems and ML infrastructure end-to-end—distributed, observable, and security-aware—while shipping production-grade AI pipelines. I want to work in teams that value reliability, modular design, and measurable impact.

I’m a backend & ML infrastructure-focused builder who turns research and engineering into systems people can actually run. I’ve developed SecureX, a developer API security platform that centralizes authentication, traffic filtering, and threat analysis—deployable in minutes—and built a history-driven risk scoring engine with a fail-closed security model.

I also design AI pipelines and local-first tools with production thinking: an 8-stage Generative Production Pipeline that automates short-form video creation with checkpoint recovery at every stage, and a privacy-first desktop productivity tracker in Rust where the Rust core operates independently and an optional Dockerized LLM summarizer (Ollama) adds value. I care about reliability, modularity, and observability—whether it’s fault-tolerant distributed architecture, unified observability APIs, or instrumentation via Prometheus and Grafana on my home lab services.

Experience

Work history, roles, and key accomplishments

Generative Short-Video Pipeline

AI Short Video Factory

Automated an 8-stage generative production pipeline—from topic selection to YouTube upload—creating fully edited short-form videos with zero human intervention. Implemented checkpoint recovery at every stage and built a modular stage-isolated architecture with independently swappable LLM, diffusion, and TTS components.

Python Diffusion Models Piper TTS FFMPEG Modular Pipeline Design

SecureX API Security

SecureX

Built a centralized API security platform that eliminated 2–3 hours of manual protection work per developer by providing authentication, traffic filtering, and threat analysis in one deployable system. Engineered a history-driven risk scoring engine and fault-tolerant distributed architecture with unified observability and fail-closed request blocking by default.

Python fastAPI Redis Docker Distributed Systems Fault Tolerance Observability Threat Detection

GnomCTL Wayland Telemetry

GnomCTL

Developed a Wayland-native window telemetry solution to overcome Wayland’s security model that broke existing window-tracking tools. Integrated as a native GNOME Shell extension that emits structured focus telemetry via a single CLI call with no screenshots, OCR, or network calls.

GNOME Shell IPC Wayland CLI

Local-First Productivity Tracker

Offline Productivity Tracker

Built a privacy-first desktop activity tracker in Rust that logs window focus, terminal commands, and git activity as append-only JSONL with zero screenshots, zero OCR, and zero network egress. Added an optional Dockerized Ollama summarization module for daily work summaries while keeping the Rust core independently testable with 20 unit tests.

Rust Python Docker Local First Architecture Data Enrichment Unit Testing

Home Lab Production Server

Home Lab

Operated a self-managed bare-metal Linux server running production services with SSL termination, Traefik + Nginx subdomain routing, and secure remote access via Cloudflare Tunnels. Implemented observability with Prometheus and Grafana and orchestrated deployments using Docker Swarm with rolling updates, health checks, and automatic container recovery for live SecureX and ML experiment endpoints.