Prasoon Parashar
Solving real-world problems with code, coffee, and just the right amount of overfitting.
I believe the best machine learning solutions don't just work — it makes life better. Whether it’s optimizing healthcare models or building tools that actually get used, I focus on making AI helpful, not just accurate.
And when I’m not debugging models, I’m probably fine-tuning my Spotify playlists or climbing the Valorant ranks (one tense match at a time)
About Me
Hey — I’m Prasoon, a Machine Learning Engineer who enjoys building things that make a difference. Whether it's fine-tuning LLMs for clinical insights, designing voice assistants for pediatric care, or convincing pipelines to behave (most days), I love solving real-world problems with data and AI.
My journey has taken me through research labs, startups, and enterprise teams — and each experience has taught me that impactful ML isn’t just about accuracy; it’s about usefulness. I’m especially drawn to projects in healthcare, accessibility, and anything that makes life a little easier for someone out there.
In the long run, I want to keep bridging AI with human-centered design — and build something that genuinely makes people say: “Huh, that’s actually helpful.” If you have anything in mind, lets have a chat!
Experience
Cohere Health - Machine Learning Engineer Co-op
Boston, MA | January 2025 - May 2025
At Cohere, I worked on building smarter clinical AI systems that actually make a difference. I fine-tuned models like Phi-2 and LLaMA-3.1 to extract meaningful relations from medical notes — but the real win was designing a graph-based sampling pipeline that boosted both training efficiency and F1 scores significantly. I also had fun (yes, fun) reducing deployment time by 40% through adapter optimization in SageMaker. Collaborating closely with medical experts taught me how vital explainability is when building AI for real people, not just metrics.
Stevens Institute of Technology - AI Research Intern
Hoboken, NJ | May 2024 - August 2024
My summer research revolved around getting LLMs (like LLaMA-3) to do a better job at persuasion detection — yes, we made AI slightly more convincing. I experimented with prompting techniques like self-consistency and few-shot examples, improving precision by 20% and cutting down hallucination. I also built a repeatable evaluation pipeline to test how prompt variations affected performance.
Macquarie - Data Scientist
Gurgaon, IND | January 2023 - August 2023
At Macquarie, I turned raw financial data into something more insightful and far less manual. One of my key contributions was building a Python automation tool using Pandas and NumPy to compare taxonomy returns — a process that was previously slow and error-prone. The tool streamlined report reconciliation and slashed turnaround time by 60%. I also improved pipeline efficiency using BigQuery-based clustering and created dashboards that helped teams visualize trends. Mentoring other interns along the way was an added bonus — and one of the most rewarding parts of the experience.
Think Future Technologies - Machine Learning Intern
Gurgaon, IND | May 2022 - August 2023
During this internship, I worked on making speech recognition smarter — and more understandable. I built a post-processing module using NLP techniques (like NLTK and spaCy) that cleaned up transcription outputs and boosted overall accuracy by 20%. On the computer vision side, I created a heuristic-based feature selection method using OpenCV, which sped up object detection by 20% without sacrificing accuracy. I also hand-tagged and curated a dataset to train a high-precision NER model — the kind of grunt work that taught me a lot about how quality data really powers good ML.
Education
Stevens Institute of Technology
Master of Science in Computer Science
August 2023 - May 2025
CGPA : 4.0 / 4.0
Netaji Subhas University of Technology
Bachelor of Technology in Computer Engineering
August 2019 - May 2023
CGPA : 3.6 / 4.0
Certifications
AWS Machine Learning Enginner: Associate
June 2025
AWS AI Practitioner: Foundational
May 2025
Databricks Apache Spark Developer Associate Certification
August 2024
Get in Touch
I'm passionate about leveraging machine learning to drive business growth and innovation. Whether you have a project in mind or just want to connect, I would love to hear from you.