Snehal Nair

Senior AI Engineering Leader | Head of AI & Innovations

Edinburgh, UK | snehal.1409@gmail.com | Linkedin | Medium | Visit My Website

Executive Summary

Senior ML and AI leader with 10+ years building production AI systems and the teams that deliver them. Built Viator’s AI function from zero to 12+ Principal and Senior Scientists and Engineers; owned the roadmap, established governance, and delivered GBP 50M+ in measurable commercial impact across search, personalisation, GenAI content pipelines, and agentic AI. Published at ACM KDD 2025 on cost-aware LLM optimisation. Deep hands-on experience across the full modern AI stack – LLM fine-tuning, RAG and GraphRAG, agentic systems, trajectory-level evaluation, and production MLOps. Comfortable setting technical direction, translating AI complexity for non-technical stakeholders, and maintaining engineering rigour through delivery.

Core Competencies

  • People Leadership: Building and scaling ML teams (0 to 12+); 1:1s and career development; coaching senior contributors; hiring and onboarding; OKR facilitation
  • Generative AI and LLM Systems: LLM fine-tuning (SFT, RLHF-adjacent); RAG and GraphRAG; prompt optimisation (ACM KDD 2025); agentic AI with trajectory-level evaluation; cost-aware inference; 27% cost reduction in production
  • Applied ML: Learning-to-Rank; real-time personalisation; multilingual NLP (ABSA, 20+ languages); churn propensity; energy disaggregation (NILM); deep learning (CNN-RNN, LSTM)
  • MLOps and Infrastructure: CI/CD eval integration; model monitoring (Arize, Grafana); Kubernetes, Seldon, Valohai, Airflow; AWS (SageMaker, EMR, Kinesis); PySpark at batch scale
  • Responsible AI: Safety gate architecture; adversarial red-teaming; PII policy; GDPR-aligned data handling; co-designed GenAI safety programme with University of Edinburgh
  • Stakeholder Communication: C-suite and board-level engagement; translating ML complexity into commercial decisions; cross-functional alignment across product, engineering, legal, and commercial teams

Professional Experience

Viator (Tripadvisor Group) | London, UK

Machine Learning Manager (Head of AI function) | Mar 2022 - Dec 2025

  • AI Function Inception: Built and scaled the AI function from 0 to 12+ Principal and Senior Scientists, establishing an “Ablation-First” culture that justified every deployment via ROI and cost-performance.
  • Strategic Impact: Delivered 50M+ in business impact through a suite of AI products, including Search Optimization, Personalized Recommendations, and Image Enhancements.
  • GenAI & LLMOps: Built policy-compliant LLM infrastructure (AWS, TogetherAI) with traffic control and observability, reducing cost-per-inference by 27%.
  • Responsible AI Governance: Designed a “Critic-at-the-Edge” Safety Gate that reduced high-risk incidents by 92% with <10ms validation latency.
  • Thought Leadership: Lead author for ACM KDD 2025 research on cost-aware prompt optimization; co-designed GenAI research programs with the University of Edinburgh.

Scottish Power | Glasgow, UK

Lead Data Scientist | Oct 2019 - Mar 2022

  • Predictive Analytics: Led development of statistical and ML models for energy forecasting, customer churn (propensity), and NILM-based energy disaggregation.
  • Capability Building: Founded the internal “ML Academy” to accelerate AI literacy and cross-organizational adoption of advanced analytics.

Queryclick | Edinburgh, UK

Senior Data Scientist | Sep 2019 - Sep 2020

  • High-Scale Modeling: Built deep learning attribution models (CNN-RNN, LSTM) processing 4M+ daily events; improved attribution accuracy by 30% post-cookie depreciation.

Nielsen | Dubai, UAE

Senior Data Science Manager | Jul 2012 - Oct 2016

  • Regional Leadership: Led a 15-member team delivering ML solutions in telecom, retail, and media across the Middle East and North Africa.
  • Operational Excellence: Reduced project delivery timelines from 90 days to 5 days via workflow automation; recognized with the “Excellence in Leadership Award”.

Key Technical Projects

  • Cost-Aware Automatic Prompt Optimization (APO): Built a hybrid APE-OPRO architecture that matched SOTA performance while reducing API costs by 18% and optimization time by 33%.
  • Real-Time Personalized Search (Ranker V2): Deployed a two-stage Learning-to-Rank system (LightGBM + DCN-v2) achieving a +9.5% conversion lift and reducing P95 latency to 47ms.
  • Knowledge Governance (GraphRAG): Built a self-adaptive five-stage RAG pipeline using Neo4j to manage FAQ truth propagation, reducing hallucinations by 90% and support tickets by 34%.
  • Review Summarization at Scale: Developed a multilingual ABSA and opinion clustering pipeline that reduced LLM token usage by 82% and maintained 94% theme coverage.

Technical Toolkit

  • Languages/Frameworks: Python, PyTorch, PySpark, SQL, LangChain, LangGraph, HuggingFace Transformers, BERT, DeBERTa, LightGBM.
  • Infrastructure/Ops: AWS, BigQuery, Qdrant, Redis, Arize, Valohai, Seldon (Kubernetes), Airflow, TogetherAI.

Education

  • M.Sc. Data Science, Lancaster University, UK
  • MBA (Marketing & Analytics), Lucknow University, India
  • B.Sc. Chemistry, Mumbai University, India