Jan 2025 – Present · 1 yr 2 mos · Remote · Remote

• Your friendly neighbourhood open-source AI research lab, bootstrapped.
• We read a bunch of stuff here mainly in infra, math, ml and ai and help companies for the same.
• We are a very strong team of researchers and developers from IITs, ISI, BITS, IISc, IIITs, Microsoft, Nvidia and specialize in research studies, inference engine optimization, finetuning SOTA models, Agentic AI, compute management and other things cool.
• We believe in First Principles and a lot of C++.

Jul 2024 – Jan 2025 · 6 mos · Remote · Remote

• I consult startups on their software architecture, making it more robust and secure and helping them become AI-ready. I help them set up machine learning models, mainly LLMs, and SOTA diffusion models using QLoRA and LoRA, and use the best open-source ML and AI practices to improve their existing products.
• Currently working with https://www.kreaitor.io/ as their head of product and engineering and managing a team of 20+ developers, designers & PMs. We are also a part of the Nvidia Inception and Google for Startups Program. If you're interested in cutting-edge generative AI for image generation, hit me up.
• Helped to scale it to $xxxxx MRR.

Apr 2024 – Aug 2024 · 4 mos · Remote · Remote

• Open Source Contributor at Unify AI for Generative AI use cases.

Mar 2024 – Jul 2025 · 1 yr 4 mos

• Treacle - Tiny Angel Investor
• Cybersecurity Startup incubated at IIT Kanpur
• First Angle Investment with a very tiny amount.
• 1.5X'd my money in a year and exited from Treacle in Jun 2025 because of a strategic partnership.
• HMU, if you are building any cool startups which are at a very early / pre-idea stage.

Oct 2023 – Sep 2024 · 11 mos · Remote

• Currently involved with research regarding predicting macroeconomic factors using LASSO Regression and monocasting under Prof. Ankit Kumar, Economics Dept. at IIM Calcutta. I am also working on a project with the Ministry of Finance with my professor.

Jun 2023 – Jan 2025 · 1 yr 7 mos · Pennsylvania, United States · Remote

• Head of Engineering at Invsto. Invsto is an Investments super app (inc. tooling and infrastructure) and Quantitative Sciences firm for LFT algorithms. I design algorithmic trading systems end to end and manage all forms of engineering for the company. We have received close to ~$200k in grants.
• Made fast retail trading systems on non specialized hardware optimizing on throughput and cost.
• Check out invsto.com and sharksigma.com

Jul 2022 – Dec 2024 · 2 yrs 5 mos · Bengaluru, Karnataka, India · Hybrid

• CPlat - IDC (Sigma)
• I worked with Bluetooth, Operating Systems and Hypervisors in C++ and led the bluetooth latte project in India. I also led the integration between Universal Print and M365 Web, iOS & Android applications :)
• Directly was looking after a team of 5 engineers as the senior most engineer in the team.
• Took technical design interviews across all roles at Microsoft from Principle to Junior Engineers.
• I was also contributing to Microsoft's AI school initiative, where I taught LLM theory and the math behind it.
• Left to pursue personal goals :)

Aug 2020 – Jun 2022 · 1 yr 10 mos · Bengaluru, Karnataka, India

• I worked in the product quality team of Marketplaces Flipkart, India’s largest e-commerce firm part of the Walmart Group and have worked on HLD and LLD of complex microservices. Besides this, I worked on Kubernetes and serve as the SCRUM master for the team. Among the projects I have worked on, the following are the significant features:
• Actioning service: Developed a service where any brand or vertical regulation changes will automatically delist and relist products on Flipkart from scratch. This also helped save 40 hours/week of ops bandwidth and worked on both the low and high-level design and implementation for it.
• Juno: Before sellers enter into the Flipkart ecosystem, sellers need to get approvals under certain brands and verticals, Juno as a system handles all of that. Juno is a combination of 10 microservices following an orchestrator pattern architecturally. I work on the LLD and design for most of the apps in JUNO.
• Expiry Workflow: Designed an Activiti(BPMN) based workflow and implemented it to make non-expirable invoices expirable in nature, reducing the number of fakes by 10 percent and increasing the seller NPS by 20 percent.
• Kubernetes Stateful Onboarding: Did the Proof of Concept for onboarding a stateful service; in this case, the leading team owned Elastic Search server to Kubernetes.
• OTC Location Approval Flow: Location and listing level actioning of items by using ops ingestion pipelines to notify fulfilment systems and provide the necessary order cancellations for a particular listing being sold at a particular location.
• I also am an active part of campus hiring as well and have taken multiple problem solving interviews for different roles

Jan 2020 – Jun 2020 · 5 mos · Mumbai Area, India

• Solved complex real word problems through data science and tech. The major projects were :
• Banking chatbot: Many banking and finance firms try to provide 24*7 service to their customers and do that; NLP is required. We used IBM Watson and deployed it on a nodeJS server for its easier NLP processing and configurable dialogue flow.
• Image Processing using Tesseract OCR: This project involved masking the first eight digits of the Aadhar card because of a change in government norms to lessen the manual labour. We hid it using tesseract OCR and OpenCV. In the end,60 to 70 percent accuracy was achieved on complete masking by applying grayscale and a morphological top-hat function kernel on the scaled image, saving thousands of dollars.
• Collateral Tracker: Came up with a web scraping solution using google news and magicBricks APIs to see which companies laid off because of covid 19 and how the loans sanctioned based on the salaries of those fired will be affected. We also checked how property rates would be affected because of covid-19 and which properties can’t be used as sufficient collateral for the loans already sanctioned by Piramal

May 2018 – Jul 2018 · 2 mos · Lucknow Area, India

• Worked with senior police and cyber-security officials to solution for significant problems in the police dispatch system on the following issues:
• Reduction in PRV dispatch time: Used PERT chart techniques to track the bottleneck in the dispatch of police relief vehicles in case of a distress call and made a python application to give the shortest path using various Graph Algorithms.
• IP Tracker: Worked on an IP tracking application using fuzzy string matching to avoid DDOS attacks.
• HRMS: Made significant improvements to the existing Human Resource Management software for lower police officials and solved a systematic GANTT chart approach for the rigorous on-call roster.