Welcome to My Portfolio

Welcome to My Portfolio

1

<!--

df.info()

-->

Hi!
I'm Nishith

A Data Science graduate from Northeastern University and AWS Certified Cloud Practitioner, I specialize in designing and deploying end-to-end Generative AI and RAG architectures that drive real-world impact.

/////////////////////////

/////////////////////////

/////////////////////////

/////////////////////////

/////////////////////////

/////////////////////////

<!--

df.describe()

-->

About Me

About image
About image
About image
About image
About image
About image
About image
About image

Follow me:

Social Icon
Social Icon
Social Icon
I’m a Data Scientist and Machine Learning Engineer based in Boston, driven by the challenge of turning messy, unstructured data into systems that work in the real world.
My journey into AI has always been shaped by curiosity: how far can we push models, and what does it take to make them reliable, fast, and production-ready? Over the past few years, I’ve moved far beyond basic analysis to build scalable Generative AI systems and serverless RAG pipelines.
I thrive at the intersection of AI research and real-world engineering. Whether I'm architecting LLM agents capable of handling tens of thousands of queries per second or optimizing multimodal video workflows for edge devices, my focus is always on the practical problems that decide whether a model succeeds in production:
● Reducing inference latency ● Slashing compute costs ● Scaling infrastructure without breaking reliability ● Orchestrating complex pipelines with MLOps best practices
At my core, I love building tools that help people make sense of information. To me, AI is more than just models; it’s engineering, architecture, and thoughtful design and that’s the space where I feel most at home.
About image
About image
About image
About image

Follow me:

Social Icon
Social Icon

<!--

Career Timeline

-->

Experience

Crewasis

Data Science Co-Op Jan 2025 - Jun 2025

● Architected a scalable multimodal video analysis pipeline transforming unstructured video into structured insights via frame-level ORB feature extraction and transcript analysis; orchestrated workflows using LangChain and Amazon Bedrock. ● Optimized processing latency by 300-400% and slashed per-frame analysis costs by 50% by implementing batch processing and identifying key frames with computer vision techniques, storing artifacts in Amazon S3 for downstream tasks. ● Benchmarked and tuned LLM pipelines with RAGAS, identifying the optimal prompting and chunking setup for reliable retrieval. ● Implemented robust observability with LangSmith, enabling real-time monitoring of trace latency and token usage during the development lifecycle. ● Delivered a high-performance multimodal video analysis tool that helped secure client contracts worth $60K–$180K, enabling data-driven insights from public video content.

Crewasis

Data Science Co-Op Jan 2025 - Jun 2025

● Architected a scalable multimodal video analysis pipeline transforming unstructured video into structured insights via frame-level ORB feature extraction and transcript analysis; orchestrated workflows using LangChain and Amazon Bedrock. ● Optimized processing latency by 300-400% and slashed per-frame analysis costs by 50% by implementing batch processing and identifying key frames with computer vision techniques, storing artifacts in Amazon S3 for downstream tasks. ● Benchmarked and tuned LLM pipelines with RAGAS, identifying the optimal prompting and chunking setup for reliable retrieval. ● Implemented robust observability with LangSmith, enabling real-time monitoring of trace latency and token usage during the development lifecycle. ● Delivered a high-performance multimodal video analysis tool that helped secure client contracts worth $60K–$180K, enabling data-driven insights from public video content.

Crewasis

Data Science Co-Op Jan 2025 - Jun 2025

● Architected a scalable multimodal video analysis pipeline transforming unstructured video into structured insights via frame-level ORB feature extraction and transcript analysis; orchestrated workflows using LangChain and Amazon Bedrock. ● Optimized processing latency by 300-400% and slashed per-frame analysis costs by 50% by implementing batch processing and identifying key frames with computer vision techniques, storing artifacts in Amazon S3 for downstream tasks. ● Benchmarked and tuned LLM pipelines with RAGAS, identifying the optimal prompting and chunking setup for reliable retrieval. ● Implemented robust observability with LangSmith, enabling real-time monitoring of trace latency and token usage during the development lifecycle. ● Delivered a high-performance multimodal video analysis tool that helped secure client contracts worth $60K–$180K, enabling data-driven insights from public video content.

Humanitarian Technology (HuT) Labs

Undergraduate Researcher Jul 2021 - Apr 2023

● Engineered a custom U-Net architecture for an autonomous tank-cleaning robot, implementing semantic segmentation to detect dirt on variable surfaces; achieved a 21% accuracy improvement over base baselines in both training and testing. ● Developed a privacy-first, offline accessibility app using Flutter, integrating COCO-SSD and OCR for real-time, on-device inference, specifically optimized for low-resource edge environments. ● Published 2 IEEE conference papers and led technical onboarding for junior researchers, driving knowledge transfer on computer vision pipelines and mobile deployment.

Humanitarian Technology (HuT) Labs

Undergraduate Researcher Jul 2021 - Apr 2023

● Engineered a custom U-Net architecture for an autonomous tank-cleaning robot, implementing semantic segmentation to detect dirt on variable surfaces; achieved a 21% accuracy improvement over base baselines in both training and testing. ● Developed a privacy-first, offline accessibility app using Flutter, integrating COCO-SSD and OCR for real-time, on-device inference, specifically optimized for low-resource edge environments. ● Published 2 IEEE conference papers and led technical onboarding for junior researchers, driving knowledge transfer on computer vision pipelines and mobile deployment.

Humanitarian Technology (HuT) Labs

Undergraduate Researcher Jul 2021 - Apr 2023

● Engineered a custom U-Net architecture for an autonomous tank-cleaning robot, implementing semantic segmentation to detect dirt on variable surfaces; achieved a 21% accuracy improvement over base baselines in both training and testing. ● Developed a privacy-first, offline accessibility app using Flutter, integrating COCO-SSD and OCR for real-time, on-device inference, specifically optimized for low-resource edge environments. ● Published 2 IEEE conference papers and led technical onboarding for junior researchers, driving knowledge transfer on computer vision pipelines and mobile deployment.

The Sparks Foundation

Data Science & Business Analytics Intern Mar 2022 - Apr 2022

● Created a multi-filter interactive Tableau dashboard (month, year, and country) to analyze COVID-19 dynamics across BRICS nations, visualizing case trends, testing rates, and vaccination distributions. ● Applied user-centered design principles to create intuitive visual layouts, helping users quickly identify trends in case growth, testing rates, and vaccinated vs. unvaccinated population segments.

The Sparks Foundation

Data Science & Business Analytics Intern Mar 2022 - Apr 2022

● Created a multi-filter interactive Tableau dashboard (month, year, and country) to analyze COVID-19 dynamics across BRICS nations, visualizing case trends, testing rates, and vaccination distributions. ● Applied user-centered design principles to create intuitive visual layouts, helping users quickly identify trends in case growth, testing rates, and vaccinated vs. unvaccinated population segments.

The Sparks Foundation

Data Science & Business Analytics Intern Mar 2022 - Apr 2022

● Created a multi-filter interactive Tableau dashboard (month, year, and country) to analyze COVID-19 dynamics across BRICS nations, visualizing case trends, testing rates, and vaccination distributions. ● Applied user-centered design principles to create intuitive visual layouts, helping users quickly identify trends in case growth, testing rates, and vaccinated vs. unvaccinated population segments.

<!--

Skills

-->

Languages

Python

Python

Python

C++

C++

C++

SQL

SQL

SQL

Bash

Bash

Bash

Databases

MySQL

MySQL

MySQL

PostgreSQL

PostgreSQL

PostgreSQL

MongoDB

MongoDB

MongoDB

ChromaDB

ChromaDB

ChromaDB

FAISS

FAISS

FAISS

Frameworks

Generative AI and LLM

LangGraph

LangGraph

LangGraph

LangChain

LangChain

LangChain

Hugging Face

Hugging Face

Hugging Face

Transformers

Transformers

Transformers

FastMCP

FastMCP

FastMCP

RAGAs

RAGAs

RAGAs

LangFuse

LangFuse

LangFuse

LangSmith

LangSmith

LangSmith

Machine Learning and Deep Learning

TensorFlow

TensorFlow

TensorFlow

Scikit-learn

Scikit-learn

Scikit-learn

NLTK

NLTK

NLTK

Pandas

Pandas

Pandas

Numpy

Numpy

Numpy

OpenCV

OpenCV

OpenCV

PySpark

PySpark

PySpark

MLOps

MLflow

MLflow

MLflow

Airflow

Airflow

Airflow

DVC

DVC

DVC

Docker

Docker

Docker

Kubernetes

Kubernetes

Kubernetes

Git

Git

Git

GitHub Actions

GitHub Actions

GitHub Actions

Cloud

Amazon Web Services - Bedrock, S3, EC2, Lambda, Glue, RDS, DynamoDB

Amazon Web Services - Bedrock, S3, EC2, Lambda, Glue, RDS, DynamoDB

Amazon Web Services - Bedrock, S3, EC2, Lambda, Glue, RDS, DynamoDB

Google Cloud Platform - Cloud Run, Eventarc, Cloud SQL, Cloud Storage

Google Cloud Platform - Cloud Run, Eventarc, Cloud SQL, Cloud Storage

Google Cloud Platform - Cloud Run, Eventarc, Cloud SQL, Cloud Storage

<!--

Certifications and Others

-->

AWS Certified Cloud Practitioner

2025

AWS Certified Cloud Practitioner

2025

AWS Certified Cloud Practitioner

2025

Reviewer for ICRM 2025

2025

Reviewer for ICRM 2025

2025

Reviewer for ICRM 2025

2025

Vice President at Khoury Master Students Council

2025

Vice President at Khoury Master Students Council

2025

Vice President at Khoury Master Students Council

2025

Selected for Best Project in IE7374 MLOps course at NEU

2024

Selected for Best Project in IE7374 MLOps course at NEU

2024

Selected for Best Project in IE7374 MLOps course at NEU

2024

Graphic Designer at Khoury Master Students Council

2024

Graphic Designer at Khoury Master Students Council

2024

Graphic Designer at Khoury Master Students Council

2024

/////////////////////////

/////////////////////////

/////////////////////////

/////////////////////////

/////////////////////////

/////////////////////////

<!--

df.query()

-->

Let’s Work Together

Artificial intelligence is the future, and the future is here. - Fei Fei Li

Social Icon
Social Icon

Artificial intelligence is the future, and the future is here. - Fei Fei Li

Social Icon
Social Icon

Artificial intelligence is the future, and the future is here. - Fei Fei Li

Social Icon
Social Icon

Create a free website with Framer, the website builder loved by startups, designers and agencies.