Available for remote roles — immediate start

AI / ML Engineer

Alan Monson
Chacko

LangChain · RAG · XGBoost · FastAPI · LangGraph

I build production-grade AI systems — not demos. From gradient-boosted propensity models to enterprise RAG chatbots, my work is live, measurable, and in production.

0.0000
ROC-AUC · Sale propensity model
0%
QA pass rate · RAG chatbot
0%
Runtime crashes · NLP pipeline
0k
Land Registry records processed
View my work GitHub ↗ LinkedIn ↗
Scroll

What I've shipped

Production Projects

🏠

UK Real Estate Sale Propensity Platform

End-to-end ML pipeline predicting residential property sale likelihood within 12 months. Ensemble of XGBoost, LightGBM, CatBoost with Optuna Bayesian hyperparameter optimisation. SHAP explainability served via FastAPI in under 100ms.

↑ ROC-AUC 0.8628 · 149k transactions · <10s training · 0% data leakage
XGBoostLightGBMCatBoost OptunaSHAPFastAPI SQLitescikit-learn
View on GitHub →
🤖

Enterprise RAG Conversational Chatbot

Production RAG system querying 42 proprietary documents (452 semantic chunks) for 24/7 automated enterprise Q&A. BAAI/bge-m3 1024-dim local embeddings, history-aware pre-retriever, and automated citation compiler.

↑ QA pass rate 76.9% → 100% · 0% hallucinations · 100% data privacy
LangChainChromaDBHuggingFace OpenAI APIStreamlitGradio
View on GitHub →
🏷️

LLM Semantic NLP Tagging Engine

Production NLP pipeline extracting 50+ structured property tags from unstructured UK real estate listings. Type-safe Pydantic schemas output binary feature flags with citation strings. Includes TP/FP/FN evaluation dashboard and live API cost tracking.

↑ 50+ tags · 0% runtime crashes · Real-time cost monitoring
LangChainPydanticGPT-4o-mini Gemini 2.0 FlashPython
View on GitHub →
🔍

AI Job Search Automation Tool

4-stage agentic pipeline powered by Claude API. Covers ATS keyword scoring, resume tailoring with JD-matched bullets, personalised cover letter generation, and mock interview Q&A graded 1–10 with improvement feedback.

↑ 4 agentic stages · Structured JSON outputs · Real-time grading
Claude APIReactAnthropic Prompt EngineeringStructured JSON
View on GitHub →
⚙️

AutoML Framework

Configurable end-to-end ML pipeline covering classification, regression, and clustering. Automated preprocessing, model selection, SHAP-based explainability, and PDF diagnostic report generation. Tested on student performance, wine quality, and car pricing datasets.

↑ 3 task types · SHAP reports · Auto PDF diagnostics
scikit-learnSHAPRandom Forest SVMPython
View on GitHub →
🕸️

Multi-Agent AI Automation Workflows

Agentic sales research system using Relevance AI (Company + Prospect + Report tools) for automated pre-call briefings. n8n pipelines for AI content generation and WhatsApp delivery. 24/7 agentic chatbot using LangGraph for stateful workflow management.

↑ Multi-agent orchestration · Real-time web data · LangGraph stateful
n8nRelevance AILangGraph LangChainGemini
View on GitHub →

What I work with

Technical Skills

LLM & Agents
LangChainLangGraph OpenAI APIGemini 2.0 HuggingFaceClaude API Prompt Engineering
RAG & Vector DB
ChromaDBBAAI/bge-m3 Semantic chunkingMulti-turn retrieval Citation pipelines
ML & Ensemble
XGBoostLightGBM CatBoostOptuna SHAPscikit-learn Random ForestSVM
Backend & APIs
FastAPIPython PydanticSQLite REST APIsStreamlit Gradio
Cloud & Data Eng
AWS S3/Lambda/GlueAzure ADF DatabricksApache Spark Delta LakePySpark
Computer Vision
YOLOOpenCV TensorFlowKeras

Where I've worked

Experience

2025 – Present
Technical Intern — AI, ML & RAG Systems
PropMarker · Remote (UK client)
  • Deployed gradient-boosted sale propensity model (XGBoost, LightGBM, CatBoost) achieving ROC-AUC 0.8628 on 149k Land Registry records
  • Built production RAG chatbot (42 docs, 452 chunks) lifting QA pass rate from 76.9% → 100% with zero hallucinations
  • Engineered NLP tagging engine extracting 50+ structured tags with 0% runtime crashes in continuous production
  • Deployed SHAP explainability via FastAPI computing feature contributions in under 100ms per prediction
Sep 2025 – Present
ML Engineer Intern — AutoML Framework
DataKompany · Remote
  • Engineered configurable AutoML pipeline covering classification, regression, and clustering with automated SHAP explainability
  • Generated automated PDF diagnostic reports with model persistence for direct production deployment

Get in touch

Let's build something

I'm open to remote full-time roles, contracts, and paid internships in AI/ML engineering. Available immediately.