May 2022 – Aug 2024 · 2 yrs 3 mos · Bengaluru, Karnataka, India

• Wrote Order Allocation module for on-demand logistics system facilitating foods/goods deliveries to Minimize ETA of orders & wait time of drivers, with uniform order allocation among the pool of drivers.
• Created LLM-based (using GPT3.5 turbo,llama-2 7b, Mistral 7-b) non-deterministic chatbot with Retrieval and Parallel API calling capabilities, potentially saving 50-70k USD per month.
• Created Data Catalogue Tool for the Data Engineering \& Analysis team for better searches, documentation, and data lineage graphs using multiple auto-generative LLM(Mistral 8x7b) bot-network.
• Worked on Abuse and theft maintenance model of bikes, geographically indexing area for geofencing using H3 index, and capable of detecting falling off the bike with 100\% accuracy.
• Worked on battery demand prediction model, adjusting the battery numbers based on truck arrivals and traffic patterns, reducing RMSE from 20 to 2.2.
• Developed a Voltage Smoothening Filter in Dumb (1.x) batteries, decreasing the false positive case of Long-term-rental bike swap by 65\%.
• Developed Wrong Way Detection for flagging out vehicles moving in on opposite street direction Patent No. 202241059204.
• Worked on Hex2Vec Model, similar to Word2Vec, converting Uber H3 geographical index to vector embedding and using those feature vectors to create classifier model to determine new Yulu Zones, with PR AUC 0.83 .
• Solved Dynamic-Last mile delivery problem for Battery Supplies and scheduling, via OR-based Truck routing with Time window \& varying load, reducing time from 3 hours to 1.4 hours for single loop.

Mar 2020 – Dec 2020 · 9 mos

• Working on Photonic Ising machine for financial problems.
• Worked on LIBOR -interest rate derivatives pricing using IBM QisKit
• Testing D-wave quantum annealer for optimization task in finance
• Proposing experimental setup for Pattern Recognition on Photonic Chip.
• Working on Credit Risk model in Quantum Finance using IBM QisKit.
• Working on CDO model of Finance and developing algorithm in IBM QisKit.
• Writing a course book on Quantum Computing and Quantum Information and it’s application

Jun 2019 – Mar 2020 · 9 mos

• Wrote B.Tech. Thesis "Quantum Computing and it's application in the field of Finance & Optimization using NISQ devices"
• Simulation of Open Quantum System using Lindblad Master Equation.
• Surveyed the Quantum Machine Learning and Quantum Finance literature.
• Proposed a Theoretical solution for classifying colour (RGB system) in Quantum field.
• Proposed 8-bit binary classification Photonic Chip.
• Worked on Quantum Generative Adversarial Network and its application in Financial service.
• Literature Survey of various papers in the field of Quantum Computing and Finance.

Mar 2020 – Apr 2022 · 2 yrs 1 mo · Tokyo, Japan · Hybrid

• Worked on Tensor Network-based Image Segmentation algorithm, which outperformed U-Net architecture in training time by \textbf{22x}.
• Implemented Temporal Fusion Transformer for Energy demand forecasting, achieving \textbf{MASE of 3.711}.
• Worked on Softbank Energy, Japan, project to develop Electricity demand forecast of combining PV installation data, to calculate excess electricity need, with MAPE of \textbf{1.44\%}.
• Worked on Short Term PV-power generation model using LSTM, achieving RMSE of \textbf{11.43}.
• Worked with Tohuku Gas for Worker Schedule problem, creating a OR-based solution, to create a weekly roaster of \textbf{10k+ employees}.
• Beta tester for newly inaugurated Advantage Beta and Hybrid Solver by D-Wave.
• Benchmarking several solvers such as Fujitsu's Digital Annealer, D-Wave Quantum Annealer, Montecarlo Solver, and Gurobi for several large-scale optimization problems.
• Scraped data from various sources such as Reddit, Open Source Maritime Portal, Twitter, etc., created a pipeline to save them and performed Proof-of-Concept correlation analysis with stock prices.
• Performed benchmarking of LSTM networks and Time-series analysis, such as ARIMA, GARCH, and VAR models,