Work Experiences
Professional ML engineering experience delivering production systems with measurable business impact across e-commerce and IoT domains.
Senior ML Engineer (Contract) at Mid-sized European E-commerce Marketplace (Client - NDA)
Aug 2022 - Nov 2025•Remote | Europe
RAG-Powered Search & Discovery
Built a production-grade RAG system that improved search-to-purchase conversion by 4% and delivered sub-500ms p99 latency at 3M queries/month
- •Led design of BM25 + dense vectors with re-ranking to define latency/error SLOs and cost targets upfront
- •Built evaluation framework (offline metrics, CI/CD regression tests, load envelopes) including scalable LLM-driven approach to create golden dataset for ranking evaluation
- •Architected for scale (3M queries/month) and delivered p99 latency under 500 ms at EUR 0.001 per query, improving search-to-purchase by 4%
Customer Lifetime Value (CLV) Prediction
Built a production CLV prediction system that improved marketing ROI by 12% through value-based customer targeting and segmentation
- •Built segmented CLV models (early-stage vs established). Improved RMSE by 15-20%, Gini 0.47 to 0.62 for budget-constrained CRM, enabling 12% higher marketing ROI
- •Diagnosed feedback loop where top-decile targeting reinforced bias; led cross-functional exploration budget experiment (10% random allocation), retrained on less-biased data for better high-potential discovery
- •Shipped weekly batch pipeline scoring 300K+ profiles with segment monitoring, costing €75/month
Real-Time Purchase Intent Scoring
Deployed a session-level purchase propensity model achieving 5% conversion uplift with 40% reduction in p99 latency through Redis optimization
- •Deployed sub-second scoring for 60K+ daily predictions (Feast + ElastiCache) with Spark Structured Streaming features (50K+ events/hour) and class-weighted LightGBM (20:1 imbalance)
- •Optimized feature fetch to achieve 40% p99 latency reduction
- •Resolved production calibration breakdown eroding marketing trust; added segment-wise reliability analysis + post-hoc calibration, restoring confidence.
LLM Generated Review Summaries
Implemented automated review summarisation using LLMs to surface key insights and sentiment for shoppers
- •Fine-tuned Mistral-7B on 100K+ reviews; cut LLM cost 60% vs GPT-4 baseline while maintaining quality
- •Partnered with product and legal to define governance standards (model cards, versioning, bias-aware retrieval, PII redaction)
- •Productionized batch RAG with automated quality checks (hallucination/toxicity/relevance)
ML Engineer at eSMART Technologies
Jun 2018 - Dec 2020•Remote | Renens, Switzerland
IoT Predictive Maintenance & Heating System Anomaly Detection
Built an anomaly detection system that reduced emergency maintenance callouts by 20% through early fault detection
- •Deployed anomaly detection + alert triage for heating systems; reduced emergency maintenance callouts by 20%.
- •Diagnosed production alert-fatigue issue causing low adoption; partnered with maintenance teams and leadership to redesign system as a prioritization tool vs. fully automated detector.
- •Evolved approach from unsupervised (residuals + LOF) to supervised as labels grew via technician human-in-the-loop validation; achieved 75% Precision@50 for high-priority alerts and restored stakeholder trust.
Building-Level Energy Forecasting & Smart Energy Advisor
Developed time-series forecasting models for building energy optimization and smart heating schedules
- •Delivered 24-hour ahead energy demand forecasting (XGBoost + weather, lag, rolling windows, holiday features) powering resident-facing recommendations; achieved under 10% MAPE and 10 percentage points increase in solar self-consumption.
- •Partnered with domain experts to design tiered cold-start strategy for newly commissioned buildings (physics-informed heuristics -> archetype models -> individualized forecasts).
- •Implemented walk-forward validation, baselines (ARIMA/Prophet), and drift monitoring.
Machine Learning Research Intern at NEC Labs America
Feb 2017 – Jul 2017•Onsite | Princeton, New Jersey, USA
Education
Master of Science in Communication SystemsEPFL (École polytechnique fédérale de Lausanne)
Aug 2015 – May 2018•Lausanne, Switzerland
- •Master Thesis: Data Analysis & Anomaly Detection in Buildings using Sensor Data
- •Built unsupervised anomaly detection system for predictive maintenance in smart buildings
- •Grade: 5.25 / 6.0
Bachelor of Engineering in Electronics and CommunicationRVCE (R.V. College of Engineering)
Aug 2007 – Jun 2011•Bangalore, India
- •Graduated with distinction - CGPA 9.23 / 10.0
Other
Signal Processing Engineer at Signalchip Innovations
Feb 2012 – May 2015•Bangalore, India
Worked on multi-user detection algorithms for WCDMA/LTE systems, contributing to 1 US patent in wireless communications signal processing
- •Joined as 5th employee of semiconductor startup, contributing to core signal processing algorithms
- •Developed and validated WCDMA uplink receiver algorithms (Path Searcher, RAKE, Multiuser Detection), achieved 3GPP NodeB receiver conformance (BER <0.001) across multi-path/interference channel conditions
- •Patent: Developed symbol level interference cancellation method (US Patent 9602240), Patent: Developed optimized channel estimation system reducing computational complexity (US Patent 20160365991)
Junior Research Fellow at Indian Institute of Science
Jun 2011 – Jan 2012•Bangalore, India
Research fellowship focused on developing efficient image reconstruction algorithms for Diffuse Optical Tomography, resulting in a published paper in Medical Physics journal
- •Developed a data-resolution matrix method (from the sensitivity/Jacobian model + regularization) to identify independent measurements. Used diagonal vs. off-diagonal structure to select informative measurements and reduce data-collection time.
- •Achieved ~20% more independent measurements than singular-value selection while preserving reconstruction quality
- •Published first-author paper in Medical Physics journal.
Work With Me
I bring hands-on experience delivering production MLOps and GenAI systems at moderate scale—with minimal infrastructure footprint and cost-effective architectures. I'm excited to collaborate on building next-generation Agentic AI systems. Whether you need expertise in MLOps, GenAI, or Agentic AI—let's connect.
Contact Me