Systems that think, interfaces that feel.

I build the layer between where AI lives and where people actually work. Not demos. Not prompts. Products that hold up when the stakes are real.

Builder with a systems bias.

I’m a computer science undergrad and software engineer working on AI-heavy products, industrial computer vision, realtime avatar systems, and private infrastructure. I’m interested in the parts that usually break after the prototype: data flow, model behavior, permissions, latency, observability, and the frontend details that make the whole thing usable.

01

I care about state, latency, failure paths, and the handoff between AI output and product decisions.

02

Most of my work sits between backend systems, product interfaces, and controlled AI orchestration.

03

I like interfaces that are calm on the surface and very deliberate underneath.

PythonGoTypeScriptReactNext.jsPostgreSQLDockerKubernetesvLLMllama.cpp

Recent work.

Roles across product engineering, AI systems, research prototypes, and infrastructure-heavy deployments.

2024 — Present

Software Engineer Trainee · Hybrid

VyomChara Pvt. Ltd.

  • Built AI case-management workflows with legal review paths and role-based approvals.
  • Developed computer-vision APIs for vehicle, personnel, and hazard detection in mining operations.
  • Worked on realtime avatar pipelines, STT/TTS orchestration, and private AI infrastructure.
May — Jul 2025

Software Engineering Intern · Dhanbad

TEXMiN, IIT ISM Dhanbad

  • Benchmarked segmentation models and built preprocessing pipelines for noisy real-world datasets.
  • Exposed model inference through REST APIs for upload, batch processing, and GPU-aware services.
  • Containerized hyperspectral microservices with memory-aware scheduling.
2023 — 2024

Project Intern · Dhanbad

TEXMiN, IIT ISM Dhanbad

  • Compared research architectures for hyperspectral mineral classification.
  • Worked with spectral camera datasets across multiple mineral classes.
  • Improved preprocessing workflows and model evaluation quality.

Things I have built.

Private investigation operations platform

01

A private AI-assisted case workflow with local model serving, human review gates, and role-based approvals.

ReactSupabaseKubernetesRAG

Live conversational avatar stack

02

A low-latency speech pipeline for a live avatar demo with memory-aware GPU scheduling.

PythonNode.jsWebSocketsTTS

Industrial edge vision network

03

Computer-vision workloads near industrial sites with lightweight telemetry back to operators.

ReactMQTTDockerCV

Hyperspectral mineral segmentation

04

Segmentation benchmarks, preprocessing pipelines, and GPU-aware inference services for spectral data.

PyTorchFastAPIDockerKubernetes

Have something worth building?

Send the rough version. I can help think through the product, the architecture, and the interface before it turns into a messy prototype.