Harekrissna Rathod

Senior Data Scientist with professional experience in causal inference, machine learning, LLM-based workflows, and large-scale experimentation in the financial sector. Designed and scaled Wells Fargo's multi-creative strategy, enabling measurement of $15M+ cumulative incremental profit value (PV) YTD — including $7.3M+ PV within the first 3 months of 2026. Deep expertise in incrementality testing, Bayesian frameworks, statistical modeling, propensity modeling, and campaign attribution.

Recognized with 1st Prize at WF Projects Showcase '24 (234 submissions), 2x Project Spotlight Awards, highlighted in 4x Line of Business Quarterly Newsletter, with findings presented to senior leadership across multiple initiatives.

Senior Data Scientist — Channel Sciences | Experimentation Lead at Wells Fargo India & Philippines (2022-07 – Present)

• Pioneered Wells Fargo's multi-creative strategy, using ~$150K pilot incremental profit to convince LOB stakeholders and scale from 2 to 21 offers (43 creatives) across CC, PL, Deposits, and Wealth & Investment Management.
• Measured $15M+ cumulative incremental profit ($150K+ to $3.6M in 15 months) by developing a novel multi-arm bandit and Bayesian opportunity-cost framework benchmarking product acquisitions against next-best eligible offer.
• Migrated to GCP; automated a Tableau MoM self-serve dashboard; reduced runtime by 80% via PySpark and Pandas parallelization; presented findings to senior leadership, featured as a star project.
• Reported $5M+ net incremental profit value via exact-matching causal attribution across Financial Health programs; presented experiment design and revenue impact at Business Conclave to senior leadership.
• Achieved a 99.2% reduction in data processing time (6+ Hours to ~3 Minutes) via a novel profile-based matching algorithm.
• Building an LLM pipeline on GCP for 35M service call transcripts – two-stage funnel (fast intent pre-filter reduces scope to ~5M actionable calls), classifying intent into Dispute/Complaint (70.3%), Product Need (22.2%), and Service Inquiry (7.5%).
• Integrating VOICE intent signals into offer arbitration system to personalize ad decisioning at scale.
• Developing a vector embedding pipeline for 1,023 ad creatives to power arbitration feature enrichment, segment-targeted creative generation, and cold-start resolution via nearest-neighbor search; queued for Q4 2026 development cycle.
• Documented 34 GCP/Enterprise Data Lake tables (500+ features spanning interaction history, product and balance data, and geo/demographic variables) as structured instruction files enabling org-wide self-service querying via GitHub Copilot.
• Built a self-serve outreach Tableau dashboard covering 17 cross-functional campaigns, reducing analysis time by 60%.
• Analyzed $10B+ in deposit outflow across 12M customers to identify key retention levers and surface tailored solutions.

Data Science Intern — Buried Threat Detection at Tata Power SED (2020-06 – 2020-08)

• Achieved real-time detection at 30 FPS with state-of-the-art accuracy; obtained 82% precision via Triplet Mining and Heated SoftMax after benchmarking Faster-RCNN against YOLO and SSD baselines.

Data Science Intern — Distribution Transformer Health Index at Torrent Power Ltd. (2021-06 – 2021-08)

• Applied ML and Fuzzy Logic to grade 2,000+ distribution transformers with 95%+ accuracy; built Deep Random Forest models to handle partially missing data; implemented a Python-based GUI for use by non-technical field operators.

B.Tech + M.Tech (Dual Degree) in Electrical Engineering – Indian Institute of Technology Bombay (IIT Bombay) (2017 – 2022) CGPA: 9.16/10