@byaarav

Aarav Kashyap Singh

I turn "can you build this?" into things that actually run in production.
I build AI systems that solve human problems. I work across backend workflows, model orchestration, product interfaces, and the details that make software feel clear to use.
I care less about what's trendy and more about what actually ships, scales, and holds up under real use.

India / remote worldwide

The work leaves a trace.

Public output, production responsibility, and the academic foundations underneath both. A record of showing up and shipping.

365-day signal

GitHub activity

Loading graph...

Experience

PythonFastAPIVision AI PipelineBalance-Chain ValidationCRM

Architected a multi-provider vision pipeline (Gemini native PDF, Claude Haiku, local Llama via Ollama) with automatic fallback to page-level extraction when native parsing fails or truncates.
Built a review-first validation model - reconciling parsed totals against statement metadata and balance movement, flagging risky extractions instead of silently trusting model output.
Designed entity normalisation and category harmonisation passes that clean up inconsistent bank-statement wording into audit-ready groupings, without altering the original printed record.

Academic journey

Education

Lovely Professional University

B.Tech Computer Science

Phagwara, Punjab

2023 - 2027

B.TechComputer Science2027 cohort

Excellence Junior College

Higher secondary education

Suchitra, Hyderabad

2021 - 2023

MathematicsPhysicsChemistry

DPS Secunderabad

Secondary education

Near Diamond Point, Secunderabad

Through 2021

School education

Difficult workflows, made dependable.

Each system has a clear path through it. Inputs are understood, decisions are checked, and the result is built to hold up in production.

LangGraph / agentic RAG / Postgres / Redis queues

AI-powered backends

From messy inputs to reliable AI workflows with clear system boundaries.

Production backends where models do real work: retrieval, reasoning, validation, and structured output.

Agentic retrieval
Guardrails + evals
Async queues
Observability + logging

LangGraph · Agentic RAG · Postgres · Redis queues

Next.js / TypeScript / FastAPI / Postgres

Full stack applications

Products that ship cleanly from schema to surface without losing coherence on the way up.

I own the full path: backend contracts, frontend interaction, deployment, and the final feel of the product.

System design
Data modeling
API design
UI delivery
Deployment
Monitoring

Next.js · TypeScript · FastAPI · Postgres

Gemini Vision / PyMuPDF / Claude / LangGraph

Document intelligence

Raw PDFs, scans, and statements turned into clean, auditable structured output.

Vision models, extraction pipelines, and validation layers built for operational use instead of demos.

Ingest
Extract
Validate
Export

Gemini Vision · PyMuPDF · Claude · LangGraph

OpenAI / Claude / tool calling / webhooks

Custom AI integrations

Language models woven into existing products so the workflow improves without the product breaking.

I integrate AI where it belongs: search, automation, and internal tools that respect the rest of the stack.

Workflow automation
Internal copilots
API tooling
Webhooks + events

OpenAI · Claude · Tool calling · Webhooks

Tools I actually use.

A working shelf of languages, models, runtimes, and interface tools that repeatedly earn their place.

Let's build something useful.

Bring the rough version: a constraint, an unfinished system, or a stubborn problem. We can make the useful path visible together.

Schedule a free call30-minute strategy session aaravkashyap1203@gmail.comQuick inquiries and questions Connect on XFollow for builds and notes

Replies within 24 hours

Open to remote, freelance, and full-time

Work hard. Play hard.