Open to opportunities

Harris Idrees

@harrisidrees

Message

I build production LLM systems, agentic RAG pipelines, and scalable AI infrastructure.

United Kingdom

Message

What I'm looking for

I'm looking for a fully remote role where I can own production AI initiatives end to end, building reliable LLM, agentic, and inference systems with measurable outcomes across cost, latency, governance, and ROI.

I've built production LLM and RAG systems for Perplexity AI, Toptal's legal intelligence client, Versana, and Anki. At Perplexity, I developed core RAG, query-routing, MCP, and LangGraph systems supporting Sonar products serving 22M+ monthly active users and 1B+ queries per month.

I've delivered multi-agent legal research platforms with live knowledge graphs, hybrid retrieval, citation grounding, and evaluation harnesses, while improving inference latency, cost governance, and model reliability. My work spans Python, LangGraph, LangChain, MCP, AWS, GCP, FastAPI, Kubernetes, and production MLOps.

Experience

Work history, roles, and key accomplishments

Current

Senior Generative AI Engineer

Current

Toptal

Sep 2023 - Present (2 years 11 months)

Designed and built a production-grade multi-agent RAG system for a legal intelligence platform, including real-time document ingestion, a live Neo4j knowledge graph, hybrid retrieval, and a six-agent LangGraph orchestration layer. Implemented evaluation harnesses and production hardening with observability and cost tracking.

Python LangGraph Neo4j RAG BM25 Cohere Cross Encoder Ragas AsyncIO Langsmith Streamlit NER Cypher Cost Optimization Observability

Senior AI Engineer

Perplexity AI

Dec 2024 - Jan 2026 (1 year 1 month)

Owned end-to-end development of core RAG pipeline components and Sonar model systems, serving over 1B queries monthly. Built query routing, MCP server integrations, agentic orchestration for Deep Research, and contributed to low-latency inference optimization and cost governance.

Python RAG Routing MCP LangGraph Prompt Caching Streaming LLM As Judge Token Budgeting Model Routing Observability Trust and Safety

Machine Learning Specialist

Versana

Jun 2020 - Sep 2023 (3 years 3 months)

Designed and maintained production-grade ELT pipelines and deployed ML models via FastAPI microservices on GCP Cloud Run. Introduced a LangChain-based NLP layer for natural-language querying and implemented model monitoring with automated retraining triggers.

Python ELT fastAPI Cloud Run Lang Chain NLP Evidently AI Model Monitoring Docker Kubernetes

Artificial Intelligence Engineer

Anki

May 2018 - Jun 2020 (2 years 1 month)

Deployed BERT and GPT-based transformer models into a recommendation engine, increasing user retention by 25%. Implemented MCP-style tool integration and applied NER and sentiment analysis to automate content tagging.

Python BERT GPT Transformers Recommendation Systems NER Sentiment Analysis MCP NLP On Device Inference

Junior Software Engineer

Code Genesis

Apr 2017 - Jan 2018 (9 months)

Built and maintained RESTful APIs for internal tools and customer-facing applications using Python/Flask, integrating with MySQL and MongoDB backends. Collaborated on code reviews, system design, and performance optimization.

Python Flask REST APIs MySQL MongoDB System Design Performance Optimization