Anand JaiswalTechnical field atlas
Résumé

AI Engineer

Anand Jaiswal.

Building backend systems and applied ML/RAG pipelines.

View my work

Projects

Selected systems with measurable constraints and inspectable implementation details.

RAG Builder

Indexed 1,174 PDF chunks into an inspectable 15.99 MB dataset for local retrieval and grounded generation.

  • Python
  • PyMuPDF
  • Sentence Transformers
  • PyTorch

FirstWeek

Indexed 46 curated sections across 5 private projects with scoped retrieval, verified citations, and 129 recorded frontend tests.

  • React
  • Express
  • FastAPI
  • LangGraph

AI PR Review Agent

Reviewed a 76-chunk repository with 4 specialist agents for $0.002247 and routed critical findings to human approval.

  • Python
  • FastAPI
  • LangGraph
  • PostgreSQL

Movie Enquirer

Searches 5,000 movies across 72,909 semantic chunks using BM25, MiniLM, and reciprocal rank fusion.

  • Python
  • BM25
  • MiniLM
  • Streamlit

MoneyPlant

Turns Telegram messages into reversible transactions through a five-minute correction window.

  • TypeScript
  • Next.js
  • PostgreSQL
  • grammY

Personal Site

Improved measured request throughput by 16.2% while reducing post-load memory by 11%.

  • Next.js
  • TypeScript
  • MongoDB
  • Vercel

Experience

Reverse-chronological roles, with the work and scope stated directly.

Jun 2026 — Present

AI Engineer

Talendy · Tech Japan

  • Building a production conversational service over LINE with LangGraph, FastAPI, and PostgreSQL-backed state.
  • Implementing RAG, multi-provider model routing, deterministic safety paths, PII redaction, Redis, Celery, and Docker.

Jan — May 2026

AI Engineer

Akatsuki AI Technologies

  • Shipped a streaming Next.js client used by about 8,200 users across 23,500 sessions.
  • Built provider-agnostic AI APIs, hybrid retrieval, authentication, analytics, integration tests, and CI security checks.

Oct — Dec 2025

Full-Stack Engineer Intern

AI Talent Force · acquired by Akatsuki

  • Built a responsive React and WebRTC client for a real-time AI interview product.
  • Implemented a GPU-backed service that converts streamed PCM audio into timestamp-aligned 30 FPS lip-synced video.

May — Jul 2025

Application Engineering Intern

Accenture

  • Delivered an Android client and Python computer-vision service for industrial gauge readings.
  • Processed 500+ documents daily at 91.6% extraction accuracy, reducing manual effort by 60–70%.

Skills

A five-second map of the tools behind the work above.

Languages
Python · TypeScript · JavaScript · SQL
ML & AI
RAG · LangGraph · Embeddings · Hybrid retrieval
Backend & Infra
FastAPI · Node.js · PostgreSQL · Redis · Docker · GCP
Tools
Git · Linux · CI/CD · WebRTC

Achievements

Selected recognition for applied engineering and machine learning.

Co-inventor · Patent publication

Wearable Non-Invasive Asthma Diagnostic System

Filed by PDPM Indian Institute of Information Technology, Design and Manufacturing, Jabalpur on 22 July 2025; published in The Patent Office Journal (India), No. 35/2025, on 29 August 2025. Application No. 202521069749 A.

View patent publication

IIT Roorkee ML Competition · Winner

ThermaOracle

Won among 120 teams with a machine-learning model achieving 3.9 RMSE, outperforming the second-best solution by 10%.

Let’s talk.

For AI engineering roles, backend systems, or technical collaboration.