Jan 2023 – Aug 2024 · 1 yr 7 mos · Gurugram, Haryana, India · On-site

• WheelsEye is a logistics startup founded in 2017 based out of Gurugram. We're a market leader in India in GPS tracking device/software (1M+ trucks, 300Cr+ Rs. ARR ) and related services like Fastag. In addition we're continuously developing & expanding our marketplace business (like Uber, but for truck booking).
• My Work:
• Led a 7-member data-science team for the Marketplace business at WheelsEye, including training/mentoring many junior team members.
• Developed multiple end-to-end ML solutions from scratch
• Problems we worked on: Shipment Price Prediction, location-based Supply/Demand estimation, Matching shippers and truckers, truck lane identification, ETA prediction, truck vehicle type classification from images
• Successfully improved accuracy of price prediction model from 55% to 73%, and for truck lanes (loading/unloading locations) from 28% earlier to 65%
• Our shipper-trucker matching model improved the "quote-percentage" (% of truckers who show interest in a demand by offering a price) from 8% to 11%.
• The work involved was both as an Individual Contributor on specific problems as well as technical management of the team, and involved constant communication with very senior product/business stakeholders including CTO/COO.

Jun 2017 – Jan 2023 · 5 yrs 7 mos · Bangalore & Gurgaon, India

• Leading a 4 member team for dynamic pricing in Flights business (Makemytrip and Goibibo):
• Multi-arm bandit models (Thompson Sampling)
• Understanding customer's price elasticity based on itinerary/flight/customer level features (Xgboost, neural-network model)
• 3rd party website (e.g. Wego/Skyscanner) pricing, International (Gulf business) pricing
• Competitor price crawling and intelligence
• Dynamic Income calculator (to identify where more revenue can be earned)
• Previously
• 1. Myra Chatbot: customer intent classification model (pretrained word/sentence embeddings), chat similarity, customer frustration score prediction (identifying which chats to transfer to agents)
• 2. Built MMT's domestic hotel recommendation system from scratch
• Problem formulation, implementation. Built data-pipelines, model-training code, and productionized the model.
• Models: LambdaMART, wide & deep neural nets, neural-collaborative filtering
• Worked with data logging teams to ensure proper logging of all the event level data (several tables of over 100 columns each)
• Helped in hiring effort in the Data Science team by recommending many top candidates from IITs (2 guys were offered job). Supervised an intern for 6 months on the hotel ranking project.

Feb 2017 – Jul 2018 · 1 yr 5 mos · Bangalore

• Worked full time on this product during Feb-May 2017, and then for another year in parallel with my data-science job at MMT
• Product: An application focused on effective group interactions, and also enables group and people discovery, with chats segmented by topics
• Responsibilities: Conceptualized and designed the product from scratch. Worked alongside a full stack engineer, and a design intern to build the app.

Jun 2016 – Feb 2017 · 8 mos · Bangalore

• Worked as a data scientist on 'Adobe Media Optimizer' product used for search engine marketing
• Computing optimal bids for ads to maximize revenue using past clicks, cost, revenue data
• Built model for the case where bids can vary with user's device type and geography
• Tools: mathematical optimization, statistical modelling and python programming

Jun 2015 – Feb 2016 · 8 mos · Mumbai Area, India

• Responsibilities: Build trading strategies with high returns, limited risk, and low turnover
• Projects: Momentum alphas, news sentiment analysis, alphas using options data (Tools: C++ )

Jun 2012 – Jul 2014 · 2 yrs 1 mo · Bengaluru Area, India

• Modelling for derivatives pricing, and for market/model risk assessment
• Fund Derivatives, and Equity Derivatives: vanilla, basket and Asian options
• Worked alongside brilliant colleagues: PhDs from Harvard/Princeton/MIT, top-ranked IIT grads etc.

May 2011 – Jul 2011 · 2 mos · Noida Area, India

• Designed a tool to help users to find attractive titles to describe their images. Received a pre-placement job offer by Adobe

May 2010 – Jul 2010 · 2 mos · Kelowna, Canada

• Worked with Prof. Heinz Bauschke, one of world's leading researchers in Optimization.
• Media: https://news.ok.ubc.ca/2010/07/20/math-whiz-from-india-completes-research-internship-at-ubc/