Sumit Singh

Senior Product Data Scientist/Analyst with 8+ years of experience delivering data-driven product decisions through causal inference, large-scale experimentation, product analytics, and machine learning across Deliveroo, Flipkart, and Mu Sigma. Experienced in designing and analyzing A/B tests, switchback and geo-randomized experiments, econometric modeling, statistical inference, and marketplace optimization. Proven track record of partnering with Product, Engineering, and Business leaders to launch high-impact products, optimize logistics, improve customer experience, and influence strategic decisions.

Hands-on expertise in Python, SQL, statistics, econometrics, hypothesis testing, and experimentation platforms. Passionate about solving complex business problems, building robust measurement frameworks, mentoring teams, and translating data into measurable business impact. Open to Senior Product Data Scientist and Senior Data/Product Analyst opportunities and available to join immediately.

DELIVEROO (Acquired by DOORDASH) - Product Data Scientist/Analyst

1. Led end-to-end experiment design and causal analysis for logistics product launches at Deliveroo's Rx-Rider Handover team, deploying switchback and bi-saturated cluster-randomized (geo-randomised) designs across a three-sided marketplace to control for network interference and spillover bias; used fixed-effects regression (PyFixest) with simulation-based power analysis to achieve MDEs below 0.1% on key logistics metrics (Rider Wait at Restaurant, Actual Order Duration).

• Designed a bi-saturated cluster-randomized experiment (80/20 treatment split across high/low saturation clusters) to evaluate a new ML order prep-time model, isolating network effects in a three-sided marketplace using custom regression specification for impact and spillover detection.

• Led a 4-hour rapid switchback experiment (ABAB/BABA sequences with displaced variants and washout buffers) to evaluate a dispatcher prioritization system that prevented ready orders from being deprioritized — achieving stat-sig improvement in late orders and cold-food outcomes while ruling out harm to guardrail metrics.

• Designed a daily alternating cluster-randomized experiment to evaluate the redesigned scheduled-order dispatch logic using adaptive release to restaurant timing, achieving stat-sig improvement in % orders delivered within the user-selected window and reducing late-food outcomes — leading to full rollout.

2. Diagnosed chronic underpowering in geo-randomized cluster experiments (~700 clusters) on core metrics (order duration, rider wait time) and drove a progressive redesign to improve power, from standard cluster-geo randomization to daily alternating and rapid switchback designs; further reduced MDEs by 20–30% by implementing CUPAC (ML predictions as pre-experiment covariates), additional metrics as covariates (on fixed effect regressions implemented via PyFixest)

3. Owned the experiment of a large cross-vertical initiative where fundamental changes to rider allocation were made. Aligned VPs across multiple verticals for the success criteria of the initiative and rolled it out following yearly gains of $18 million. Used experiment learnings to make recommendations to further improve primary metrics to the tune of $9 million

4. Replaced ad-hoc incident impact measurement with a 3-way difference-in-differences causal inference framework, enabling accurate attribution in the presence of confounders — and drove adoption across multiple teams in the Logistics space

5. Mitigated cross-experiment contamination by aligning timelines across multiple teams accounting for possible interactions between various treatment mechanisms in a complex logistics space with inter-connected systems

6. Provided technical oversight and mentorship to incoming data scientists, drove annual roadmap planning, and established code and project review standards — contributing beyond core DS responsibilities to strengthen team execution and technical rigour.

Kalinga Institute of Industrial Technology, Bhubaneswar, India

B.Tech (Computer Science & Engineering) | 2014 - 2018

CGPA: 8.2