Syed Hamza Zaidi
AI Engineer at faff
B.E. Electronics and Instrumentation Engineering, BITS Pilani
· linkedin.com/in/syed-hamza-zaidi
Agent orchestrator. I do not write code or debug manually anymore; I build agents and flows which do it for me. I have led projects in a multi-billion dollar MNC, a seed stage startup, and built software from scratch as a founding engineer.
Experience
faff is a seed stage startup backed by Nexus Venture Partners with $3 million raised, building an AI personal assistant and concierge service.
- Automated the entire customer-facing chat experience through a multi-agent system, optimizing for latency and cost.
- Worked on voice agents, WhatsApp as an interface for agents, temporal memory, RAG, and multi-agent coordination.
Founding member of the GenAI Developer Experience team.
- Designed a federated search engine for all internal knowledge using Retrieval Augmented Generation from scratch, then wrapped it in an agentic chatbot for ease of use.
- Contributed to a managed LangChain / LangGraph agent service for bring-your-own-agent workflows.
- Hosted managed MCP services on a central platform.
- Enhanced organizational transparency and trust by developing an executive dashboard with real-time visibility into organization-wide initiatives.
- Reduced monthly report generation time for the CTO from 2 days to 2 minutes by automating manual tasks across multiple teams.
- Designed a data lake in CRM Analytics and Apache Airflow pipelines to extract, transform, and load data from Splunk, Honeycomb, Salesforce, SQL databases, Postgres, BigQuery, and more to power a Tableau dashboard.
Worked in parallel to the final year of undergraduate study.
- Designed and deployed an NLP model to categorize support threads into FAQ buckets using semantic segmentation.
- Integrated the Salesforce Einstein Bot API with an internal Slack bot to provide users escalation paths in Slack support channels.
- Redesigned NLP preprocessing for Information Retrieval models, increasing the Query Resolution Rate from 2% to 16%.
- Designed and implemented an Information Retrieval model to rank documents based on Semantic Textual Similarity retrieved from multiple internal knowledge sources.
- Deployed the model to production where it served over 200 teams and 30,000+ daily active users.
- Created an automated CI testing framework for Information Retrieval models against multiple datasets.
Education
- Head Teaching Assistant for Neural Networks & Fuzzy Logic: led a team of 12 Teaching Assistants to design assignments and workshops for 120+ students.
- Courses: Information Retrieval, Neural Networks & Fuzzy Logic, Graph Mining, Data Structures & Algorithms, Object Oriented Programming, Discrete Structures for Computer Science, Digital Image Processing, Combinatorial Mathematics.
Technical Skills
Agentic AI, multi-agent orchestration, RAG, temporal memory, LangChain, LangGraph, MCP, Python, Go, FastAPI, Postgres, Redis, Docker, Kubernetes, AWS EKS, React, shadcn/ui, Apache Airflow, Tableau, CRM Analytics
| Area | Tools / Platforms |
|---|---|
| Agent systems | Multi-agent orchestration, RAG, temporal memory, MCP, voice agents |
| Backend | Python, Go, FastAPI, Flask, Postgres, Redis |
| Infrastructure | Docker, Kubernetes, AWS EKS, AWS Amplify, Apache Airflow |
| Frontend / Data | React, shadcn/ui, CRM Analytics, Tableau |
Projects
- Founding engineer at a pre-seed startup, building the initial product from scratch.
- Built a Python FastAPI backend with Redis cache using LangChain and LangGraph to design agents leveraging OpenAI, Perplexity, and Gemini models.
- Built a Go API service and Postgres database to handle user sessions, chat threads, and vectorDB.
- Deployed a horizontally scalable backend on AWS EKS as a Kubernetes cluster with Docker images.
- Built a React frontend using shadcn/ui components, served on AWS Amplify.
- Invented a novel robust authentication mechanism based on face recognition and liveness detection.
- Deployed the model via a Flask backend and a dynamic Salesforce Lightning Web Components frontend.
- Designed and implemented a custom liveness detection mechanism using real-time statistical modeling to analyze facial reflections against dynamically changing background colors, thwarting still images, videos, and deepfake attacks.