Open to opportunities

SHADAB ARIF

@shadabarif

Message

Senior AI/ML engineer building production LLM agents, RAG architectures, and MLOps for real-world impact.

Zimbabwe

Message

What I'm looking for

I want to build and deploy production LLM agent systems—RAG, tool-use, and fine-tuning—with strong MLOps, observability, and measurable impact like lower latency and faster resolution.

I’m a Senior AI/ML Engineer with 7+ years building and shipping production-grade LLM systems, multi-agent pipelines, and RAG architectures across fintech, healthcare, and enterprise verticals. I specialize in agentic workflow design with LangGraph, LLM fine-tuning (LoRA/QLoRA/RLHF), and integrating vector search, tool-use, and session memory into autonomous AI agents serving 10,000+ daily requests.

I move fast from prototype to deployment, with a full-stack mindset spanning LangChain, LlamaIndex, HuggingFace, Azure OpenAI, GPT-4, Gemini Pro, and scalable orchestration for monitoring and reliability. Recent impact includes an Arabic customer support agent that cut support resolution time by 40% across 5,000+ monthly tickets, an enterprise knowledge RAG agent that reduced employee query resolution time by 60%, and an agentic Outlook add-in that achieved 3-second latency while cutting manual email drafting by 90%. Earlier, I built risk and underwriting models with LightGBM/XGBoost and PySpark feature pipelines, including a business rule engine that reduced manual underwriting effort by 50% and improved decisioning through measurable lifts in AUC and operational outcomes.

Experience

Work history, roles, and key accomplishments

Senior AI Engineer

RVIN

Sep 2025 - Mar 2026 (6 months)

Architected a dialect-aware Arabic customer support agent using LangGraph, GPT-4o, and Pinecone/Redis, reducing support resolution time by 40% across 5,000+ monthly tickets. Built an enterprise knowledge RAG agent with conditional routing (60% faster resolution) and automated a policy-to-training-video pipeline, cutting content production time by 80%.

LangGraph GPT 4o Pinecone Redis fastAPI Azure OpenAI Langfuse WhatsApp Business API MoviePy

Freelance ML Engineer

Digital Exchange

Jul 2025 - Aug 2025 (1 month)

Designed a patient intake automation agent that extracted structured data from medical forms and routed cases, reducing intake processing time by 35% across 3 client sites. Built and deployed a LangGraph/LLM healthcare support agent handling 200+ daily queries with zero manual escalations.

LangGraph Document Understanding Retrieval With LLMs Tool Use Python fastAPI

Machine Learning Engineer

Turing

Jun 2024 - Jun 2025 (1 year)

Built an agentic RAG email assistant for Apollo Global using LangGraph, Azure OpenAI, and Azure AI Search over 2,000+ documents, deployed as an Outlook add-in with 3-second response latency and 90% less manual email drafting. Developed a multi-agent gear recommendation system improving relevance by 40% and accelerated SEC filing workflows to under 5 minutes (70% reduction).

LangGraph Azure OpenAI Azure AI Search Faiss ChromaDB EDGAR API RLHF Fine Tuning

Senior Data Scientist

Paytm

Nov 2022 - Jun 2024 (1 year 7 months)

Developed a LightGBM loan default prediction model on 2M+ records, achieving 78% AUC and reducing portfolio bad rates by 18%. Led merchant churn prediction with XGBoost/CatBoost/LightGBM on 35M+ transactions (74% AUC) and improved credit underwriting approvals by 20% while reducing bad rates by 15%.

LightGBM XGBoost CatBoost Pyspark Credit Risk Modeling AWS S3 SQL AUC Optimization

Senior Data Scientist

Protium Finance

Jan 2020 - Oct 2022 (2 years 9 months)

Built an end-to-end Business Rule Engine across 8+ credit evaluation stages, processing 5,000+ daily applications and reducing manual underwriting effort by 50%. Fine-tuned BERT for EMI default risk (76% AUC), delivered a loan amount estimation model (15% MAPE), and created stacked XGBoost/Random Forest approval models with explainability.

fastAPI Gradient Boosting XGBoost Random Forest SHAP Explainability Pyspark DBT

Data Analyst

Larsen & Toubro

Jul 2019 - Dec 2019 (5 months)

Created a real-time SQL dashboard for construction progress, reducing reporting effort by 20% and improving visibility for 10+ project managers. Trained a defect detection model on inspection imagery, achieving an AUC of 0.78 for automated structural issue flagging.

SQL Dashboards Report Automation Machine Learning