Sarthak P.

Experience

• 

  MAQ Software

  Jul 2016 - Jun 2018

  Software Engineer

  - Machine Learning: Automated support ticket classification, efforts prediction, text sentiment analysis (urgent or not), and prioritization. Was a core member of the Artificial Intelligence Task Force learned different ML concepts and conducted sessions for the newly formed team. -ETL, Data Warehousing, and BI: Developed & maintained a very complex system for revenue recognition & reporting which had 7 upstream data sources & more than a billion transactions. Was involved in developing end-to-end ETL using SSIS/TSQL. Took up a wide range of responsibilities which involved analysis, designing, development, validation/testing, conducting code reviews, running & monitoring SQL Jobs, unblocking juniors and conducting training, integrating new features regularly. -Led a team of 4 and delivered 2 data warehousing and BI-related projects within a time of 2 months. Led the development of data pipeline, reporting cubes, and reports on top of that with row-level security enabled.-Collaboration & Leadership: Took weekly seminars on new ideas and technologies for a broader audience across the locations. Was a key player in Machine Learning and AI: helped in building an AI task force by taking daily lectures and preparing problem statements. Led a team of 5 in developing an Artificial Intelligence ticket tracking system. Took conference calls with clients on requirement gathering and feature deliveries. Show less

• 

  Fractal

  Jul 2018 - Oct 2020

  -Outlier Detections: Implemented quality checks to flag potential issues in the data usingvarious outlier detection algorithms. -Ministry of Defence (NLP/NLU): Developed Entity Recognition and Entity Association models to extract domain-specific entities/phrases from reports and associates them based on the context. -Concordia - Fractal’s Product: Led AI development of the product primarily focused on research and development of the new use cases for the product. Developed and operationalized the following ML models *Entity extraction/recognition:- Extracts entities from the unstructured data and transforms into a structural form that is consumable by other layers and BI solutions. *Data classification:- Classifies products based on their descriptions. *Data unification and mastering:- Finds the final and acceptable representation for similar data rows that have different data representations. *Data mapping:- Maps the products coming from two different sources and have different textual representations. *Recommendation system:- Recommends values for blank fields based on details that are filled in for other fields by the user. *Incremental learning:- Enhances the learning of models if corrections are done by the user on the portal.Developed Python Application that communicates with other modules of the product. Show less

  • AI Engineer II

    Apr 2019 - Oct 2020

  • AI Engineer I

    Jul 2018 - Mar 2019
• 

  Thermo Fisher Scientific

  Oct 2020 - Aug 2021

  Data Science Manager

  - Data science:* Competitive Intelligence (GeM): GeM is a bidding platform. Developed an ETL to pull bidding information from this platform. Processed unstructured documents using deeplearning models. This has helped organisation to improve their participation & get sense of competition.* Desiged a recommendation engine for newly developed Mobility App. * Supply chain solution to identify potential purchasers in future to proactively ship equipment and avoid shortfalls. -Data analytics initiatives: * Google Analytics for e-commerce: 1. Evaluated different vendors 2. Provided detailed requirements and development plan.3. Conducted followup calls.4. Timely validated and reviewed progress.* Marketing dashboard. 1. Worked with vendor to move existing excel reports to Power Bi2. Developed a refreshable data model. Show less

• 

  SymphonyAI

  Aug 2021 - now

  Generative AI - Framework-Introduced Gen AI framework for rapid AI co-pilot development. It's a skill-based framework designed to addresscomplex business queries. highly configurable & easy-to-use, 5 verticals have quickly adapted within a month and deployed their AI- Co-Pilots.-Leading data science team and helping in planning and development. Coordinating between cross functional teams.-Continuously upgrading and maintainingframework. Led the developement of key features: ED, Context Prop, Ask Backs , Knowledge base etc-Currently, experimenting using LangGraphs.Retail AI Co-Pilot (Q&A):-Led the developement of a highly conversational copilot for retail vertical and successfully released to retail's end customers. Copilot is built using 10+ skills and 40+ plus tools. It's an excellent demonstration of Predictive + Generative AI. - Have set up app from scratch, defined I/Ocontracts for middle wares and developed key tools: Data Retrieval, Summary, data analyser tools, etc. -Performed weekly demos for stakeholders. And provides exceptional support during deployments & releases. - Excellent debugging skill. Helping teams in debugging and fixing issues on on-going basis. Show less - Retail Copilot POC: led team of 3 to deliver a quick POC using langchain, React Agents, and LLM.-2 hackathon winner: ISB datathon and Symphony's Natural Language to SQL- Shelf Intelligence: Developed predictive models for on-shelf availablity & product placement recommendation.- Assortment Optimisation: Led the team of 3. Built predictive models: Product Demand Transfer, Sales Forecasting, LEAP optimisation. Developed end to end feature engineering pipeline in Kunflow. Deployed and integrated models with front end application. - Validation of ML platform: Led the team of 2. Developed cross-validation framework for forecasting models. Capable of doing a comprehensive reporting using various evaluation metrics. Benchmarked anomaly detection algorithms. Evaluated customer segmentation model compared against open source offerings. Show less

  • Lead Data Scientist

    Jul 2023 - now

  • Senior Machine Learning Engineer

    Aug 2021 - Jul 2023