TalentSync

AI resume builder and ATS optimizer for job seekers. Tailored resumes, outreach, and interview prep from one profile.

Team and acknowledgments

TalentSync (job-seeker app) is a small team plus a lot of other people’s libraries.

Leadership#

Harleen Kaur leads machine learning and backend work:

  • FastAPI services and LangChain graphs
  • Model wiring and prompts
  • Job/worker path with FastStream

Tashif Ahmad Khan covers full-stack and design:

How the team works#

Frontend and backend are separate packages and containers. New LLM behavior lands behind FastAPI routes the BFF already forwards. TypeScript on the UI, Pydantic on the API. Bun and uv in CI.

Do not point agents at TalentSync-HR paths when editing this tree.

Technology stack#

Frontend

  • Next.js 16, React 18.2
  • Tailwind, Shadcn/Radix
  • Prisma, TanStack Query, Razorpay
  • Bun

Backend

  • Python >=3.13, FastAPI
  • LangChain, LangGraph
  • FastStream + Kafka
  • SQLModel/Alembic, Motor, PyMuPDF
  • uv

AI

  • Multi-provider LLM factory
  • Primary vs small model roles
  • Prompt modules under backend/app/data/prompt/

Tooling

  • Docker Compose (dev, prod, host-infra)
  • GitHub Actions deploy to a VPS

Open source thanks#

Next.js, FastAPI, PostgreSQL, Kafka, LangChain, Prisma, Tailwind, Radix, Bun, uv.

Origins#

Started as resume analysis plus a usable web app. The current split (Next.js BFF, FastAPI, Kafka workers) is what ships. The README clone URL harleenkaur28/AI-Resume-Parser is stale; use tashifkhan/TalentSync.