Ayaan Khan
ai engineer · builder
DETAIL VIEW — A·1
Activity
- Learning
- Agentic workflows in construction and real estate.
- Watching
- Beef (Show, 2025)
- Reading
- Attention Is All You Need (Vaswani et al.)
- Hardware
- MacBook Pro M4 Pro · iPhone 17 Pro · Samsung Z Fold 7
- Software
- VS Code · Claude · Codex · Gemini · Ghostty
- Supporting
- FC Barcelona
About Me
AI engineer building production LLM systems — RAG pipelines, semantic search, multi-agent workflows. Originally from New Delhi, based in Chicago. Outside the terminal, I'm an avid football fan (the real kind), play guitar, and spend too much time thinking about music and art. I think good taste is a technical skill.
Margin note — where I'm headed
After graduation in May 2026, I want to join a team where I can ship AI systems that actually matter — whether that is at a fast-moving startup or a company building infrastructure for the next wave of intelligent software. Longer term, I want to build products of my own at the intersection of AI and systems design, and eventually contribute to research that makes LLM pipelines more reliable and interpretable.
| Education Schedule | |
|---|---|
| Degree | B.S. in Artificial Intelligence |
| Minor | Minor in Architecture |
| University | Illinois Institute of Technology |
| Completion | May 2026 |
Keynote legend — coursework
Experience
The Syntec Group · Chicago, Illinois · May 2025 – Present
AI & Digital Development Intern
- 1. Built and deployed a semantic RAG chatbot on Chatbase over firm documents, delivering cited, context-grounded answers to reduce lookup time and improve response consistency.
- 2. Developed an internal agentic system using OpenAI function calling to manage building module codes through natural language, with confirmation flows for destructive operations and ChromaDB sync for semantic search.
- 3. Implemented ingestion and retrieval workflow (chunking, embeddings, indexing) across PDFs, CSVs, website pages, and WordPress blog posts, with embedding caching via Redis that reduced inference cost by approximately 65%.
- 4. Led an information architecture plus website redesign improving navigation and access to resources; used engagement analysis to iterate content performance.
Volunteers.Covihelp · Remote (India) · May 2021 – July 2021
Co-Founder
- 1. Created and managed a 24/7 helpline during India's second COVID wave, connecting thousands of patients with critical resources like oxygen, beds, and medicines.
Excelerate (Globalshala) · New Delhi, India · June 2023 – July 2023
Project Manager
- 1. Led a global team to organize an academic event with a $30,000 budget, managing documentation, risk assessment, and external outsourcing.
Projects
| No. | Project | Description | Status |
|---|---|---|---|
| 01 | Clutch ↗ rev. note: 5-stage pipeline, retries, SSE ← | Built a staging-deployed SaaS that generates research-backed courses using a multi-stage agent pipeline, orchestrating 5 stages with persisted job state, retry policies, and failure isolation to keep long-running workflows reliable. | staging |
| 02 | SyntecAgent ↗ rev. note: confirmation gates on destructive ops ← | Built an agentic Coding, Classification & Naming system using OpenAI function calling that enables users to query, add, update, and delete BIM module codes through natural language with automated sub-code assignment. | deployed |
| 03 | VeritasLayer ↗ rev. note: every claim cites its source ← | Built an evidence-traceable document intelligence platform that extracts obligations, risk signals, and structured summaries from operational documents using LLM pipelines with source-level citation. | deployed |
| 04 | Syntec AI Chatbot ↗ rev. note: dual-LLM fallback, content-hash cache ← | Shipped a semantic search and retrieval-augmented chatbot using GPT-4o and ChromaDB embeddings to deliver question answering across PDF documents, CSVs, blog posts, and website content with automated source citations. | deployed |
| 05 | InvestoChat ↗ | RAG system for real estate investment queries with multi-path retrieval (pgvector + SQL fallbacks with MMR), OCR processing, and automated table extraction from PDF brochures. | |
| 06 | Trend Analyzer for Raw Materials ↗ | Cotton price forecasting using Facebook Prophet with external regressors (oil, gas, soybeans); evaluated with MAPE. | |
| 07 | Sports and Metrics Tracker ↗ | Built an end-to-end soccer video analysis pipeline using YOLOv8 detection and ByteTrack tracking, adding temporal smoothing for stable motion analytics and CPU only local processing with robust cross-platform video input and output. |
NOTE: SELECT ANY ROW FOR DETAIL ENLARGEMENT
Systems Thinking
- SPEC 01 Ingestion & preprocessing
Clean, chunk, and index messy docs so retrieval starts from ground truth
- SPEC 02 Hybrid retrieval + reranking
Vector search alone hallucinates on structured data — combine with lexical matching and rerank
- SPEC 03 Agentic workflows & guardrails
Function-calling agents with confirmation gates before anything destructive touches production
- SPEC 04 Structured outputs & validation
Every LLM response parsed through Pydantic schemas — if it doesn't conform, it doesn't ship
- SPEC 05 Latency/cost routing
LiteLLM across models, Redis caching, and fallback LLMs — not every query needs GPT-4o
- SPEC 06 Observability & failure modes
Inngest job tracing, eval harnesses — catch breaks before users do