Prashant S.

Which parts of your work does AI handle well — and which parts still depend on you? HatAlign's first product, Alignment Resilience, is a personalized AI displacement assessment for mid-career knowledge workers. It analyzes your specific role — the tasks you actually do, not your job title — against peer-reviewed research from Science journal. Not a personality quiz. Not a chatbot opinion. A structured analysis that makes the reasoning visible. HatAlign is an alignment-first career and hiring platform. We're building clarity for both sides of hiring — starting with candidates because the ground is moving under people right now. Founding beta is open at hatalign.com.

• Launched Alignment Resilience — a personalized AI-displacement assessment that maps a user’s real-world tasks to peer‑reviewed research to show where AI can and cannot substitute work.
• Designed product research and reasoning pipeline to make assessment logic transparent (task decomposition, evidence mapping, and risk scoring).
• Built and iterated the founding beta: product roadmap, prototype prioritization, user interviews with mid‑career knowledge workers, and early customer onboarding.
• Owned go‑to‑market direction for candidate-facing product and defined roadmap items for employer-side features and integrations.
• Managed cross‑disciplinary work (ML research, product design, engineering) and prioritized features that increase clarity and trust for users.

Stealth mode startup is a company.

• Leading product and technical strategy while building an early‑stage venture in stealth.
• Defining MVP scope, conducting user and market validation, and prioritizing technical architecture for rapid iteration.
• Recruiting and coordinating early collaborators and contractors to accelerate prototype development and customer discovery.
• Owning early roadmap, tradeoffs, and go‑to‑market hypotheses while keeping IP and positioning confidential.

• Led a team of 5 scientists in the development of GenAI-powered NLP applications for the company's flagship experience management (XM) platform, utilized by over 20k customer brands.
• Directed multiple AI-powered projects through collaboration with 4+ cross-functional teams, resulting in the successful launch of 5 new AI experiences within the experience management industry.
• Developed prompt and fine-tuned large-language models (LLMs) for summarization of PDF reports, long-form video feedback transcripts, and question answering using retrieval-augmented generation (RAG).

• Founded a project to build a web-scale knowledge extraction system for retail and digital products using NLP, Graph Neural Networks (GNNs) and large language models (LLMs). Played a key role in defining research roadmap, ontology, pipeline's architecture, and ensuring data delivery to partners.
• Led a team of 3 scientists on information extraction tasks for assembling a retail product graph. Defined scientific roadmap, wrote business review documents, and oversaw data delivery efforts.
• Delivered millions of facts to Alexa knowledge base by extracting semi-structured data from 100+ websites across 5 languages at 95+ precision.
• Hired 2 full-time scientists, mentored 3 summer interns, promoted 1 Applied Scientist to a senior role, and interviewed 100+ candidates for various tech and non-tech roles at Amazon.

• Designed and developed novel, scalable information extraction techniques for detail page oriented websites to transform semi-structured data into structured knowledge and create a large product knowledge graph. The graph then enabled various downstream applications including Search, Personalization and Alexa QA through its high-quality data.
• Led the building of a web-scale knowledge extraction pipeline for constructing Alexa knowledge graph, designing and deploying a range of components: ontology development, website clustering, website filtering, information extraction, RDF data validation and filtering.
• Developed domain-agnostic, closed/open information extraction techniques for semi-structured and tabular data, improving web extraction accuracy from 60% in state of the art to 90%+ to support production requirements. Used weak and self supervision for labeling and GNNs & Transformers for better modeling.

• Analyzed Amazon.com’s browse taxonomy to quantify how structural changes propagate to shopper behavior and business metrics.
• Developed a machine‑learning model to predict the downstream impact of taxonomy edits, enabling taxonomists and ontologists to forecast outcomes before deployment.
• Built data pipelines and evaluation metrics to validate model predictions against historical changes and held‑out experiments.
• Worked closely with taxonomy stakeholders to translate model outputs into actionable guidance, increasing the team’s ability to preview effects and reduce iteration time.

At Philips, we want people to be healthy, live well and enjoy life. We are a technology company that cares about people and the planet.

• Designed and developed core search and ranking components for a product search engine, focusing on relevance and retrieval latency.
• Implemented ranking features and relevance signals to improve product discovery quality for e‑commerce scenarios.
• Built evaluation harnesses (offline relevance metrics and test queries) to measure improvements and iterate on ranking heuristics.
• Collaborated with product and engineering teams to integrate search components into the larger product pipeline and validate behavior on real queries.

• Thesis: Computational Fact Checking by Mining Knowledge Graphs
• Minor: Statistics
• Research interests:
• Machine Learning: supervised, unsupervised and semi-supervised approaches
• Relational learning: graphical models, tensor-based methods
• Social Network Analysis
• Computational Fact Checking
• Semantic Web modeling
• Information Extraction

• Successfully completed a Math foundational course and tutorials on Machine Learning and its applications in Natural Language Processing (NLP), Information Retrieval (IR), Knowledge Discovery in Social & Information Networks, and Computer Vision.
• Designed and implemented a simple 'event' search engine to search for current controversial events. This project was funded by the U.S Department of Homeland Security.
• Won Outstanding Project Award given by Yahoo! at the 2012 DSSI

• Designed and developed batch and web applications to support IT operations of Transamerica.
• Integrated an analytic tool to hedge annuity rider funds worth $22M. Led the proof of concept evaluating Spring Batch framework as a solution.
• Performed ETL operations on 5M annuity policies using Spring Batch. Wrote backend code in SQL Server to report statistics of batch loads to actuaries.

• Developed and deployed new applications for International Programs office at University of Iowa
• Worked on enhancements, integration tasks, maintenance and support for existing systems used by Office of Student Life and International Programs office.

• Created web applications for University of Iowa's students and the Office of Student Life coordinators to facilitate on-campus activities' sign-up and tracking.
• Developed an online database for University of Iowa Nursing Staff using J2EE, Stripes and Hibernate.
• Provided technical support, enhancements and maintenance to different projects.

LoginSoftTech is a company.

• Architected and developed a prototype bus e‑ticketing application for the state of Gujarat, India, covering ticket search, price comparison, booking, and ticket tracking.
• Designed end‑to‑end system architecture and implemented core backend and frontend components for a production‑facing prototype.
• Integrated payment and ticketing workflows and added features to improve user convenience and transparency (price comparison, booking status).
• Delivered a working prototype used for stakeholder demos and early usability feedback.