← all projects
>_ devarshi.dev@portfolio:~$ cat projects/allotiq█
AI / LLM

AllotIQ

Agentic IPO intelligence platform generating cited company briefs, structured analyst reports, grounded research chat, and deterministic apply/wait/avoid signals.

Project overview

AllotIQ combines a LangChain + GPT-4o-mini AI service on FastAPI with citation validation, report-structure enforcement, response caching, usage budgets, and deterministic IPO signals. Its services are deployed across Railway and Vercel with PostgreSQL and Prisma.

LangChainGPT-4o-miniFastAPIRAGPostgreSQLRailwayVercel
Verifiedresume-backed
Buildengineering focus

CASE STUDY // CONTEXT

Problem and role

Make fast-moving IPO signals inspectable enough to support a reasoned decision.

Role: Designed and shipped the AI and full-stack product surface.

SYSTEM BOUNDARIES

  • Separate model analysis from deterministic market facts.
  • Validate citations and report structure before showing research.
  • Keep usage and cache behavior bounded.

ARCHITECTURE

  1. Source capture
  2. FastAPI research services
  3. LangChain analysis
  4. Citation and structure validation
  5. PostgreSQL / Prisma
  6. Vercel and Railway delivery

OUTCOME

A cited research workflow with explicit apply, wait, and avoid signals rather than an unbounded chat response.

Open project