Vishwa Nath Jha

Building RRAM inference chip for the agent loop — weights never move, 30x more tokens/joule.

Building the full-stack sovereign superintelligence infrastructure to accelerate safe autonomous operations in regulated industries like banking, insurance, government, etc.

Redefined how Indian lending enterprises manage collections-focused customer communications with state-of-art-art purpose-built Voice AI Agents. Models for 13 Indian languages, 200M voice AI calls, 50M+ users connected, peak at 750k calls/hr at <₹2 per minute.

As ML instructor, I delivered hands-on training bootcamps for applied machine learning & deep learning framework based application development to engineers, analysts, data scientists and executive layers of data-driven organisations.

As an instructor, I enjoyed developing and delivering state-of-the-art applied coursework driven hands-on bootcamps for teaching data science and machine learning fundamentals and applied ai application design to tens of thousands of students, academic staffs & experienced industry professionals. A few of the AI-savvy organizations who collaborated with me are as follows: Fractal Analytics, Synechron, Edelweiss, GlobalEdge, Titan, L&T and more. I also collaborated with many upcoming mooc platforms like Upgrad, Zeolearn, Ufaber, etc. during this period to teach statistics, machine learning, natural language processing, deep learning and big data using scala, julia, python, R, tableau, etc.

Created an application to predict the sentiment of tweets, Facebook posts & comments. Applied various pre-processing steps to clean the data and wrote scripts to extract 60+ features. Semantic replacement of slangs, colloquial words and parsing meaning out of hash tags were key challenges in the project.

Developed algorithms for dynamic pricing and auto-classification of items into multiple categories across different online shopping platforms. It involved web crawling to extract details about millions of product items listed on different websites, data cleaning, extracting 70+ features, building and finetuning machine learning models.

Client : One of the world's largest CPG giant Description : In the beginning of each month, the team creates a shipment optimisation model customized for various geographical regions. It acted as the basis for a daily shipment plan leveraged by the execution teams. The 2 key inputs to build highly impactful and accurate predictions were: 1. Historical shipment data 2. Knowledge of regional market dynamics Also, based on actual daily shipment information, model also provided the likelyhood of achieving a monthly target and the potential gap in cases where company is over shipping or under shipping. ________________________________________________ Client : One of the world's largest CPG giant Description : Work closely with clients to create solutions to their e-commerce related business problems by : 1. Identifying and pitching new business opportunities for the client 2. Create high impact IP's for clients to measure and enhance their market presence 3. Exploring whitespaces and working to convert them into business engagements

Client : One of the world's largest search engines Description : Help them make better decisions by 1. Processing big data into condensed & consumable reports 2. Generating actionable business insights 3. Suggesting new techniques to enhance already existing ideology 4. Liaisoning between the client and the delivery team for all the adhoc requests and ongoing projects 5. Engaging with the client to understand business dynamics and identify questions that can be addressed through analytics 6. Managing operational workflow and ensure seamless delivery for both quick turnaround analysis and long term projects 7. Identifying and harvesting new business opportunities at the client side by preparing business plans