At GrantX, I'm building the core AI infrastructure powering grant matching for thousands of organizations. Currently architecting GRASP (Grant Retrieval & Assessment System), a next-generation multi-layered intelligence platform that combines hybrid retrieval methods with chain-of-thought reasoning, multi-dimensional scoring, and evidence-based verification to ensure precise grant-opportunity matching. I architected and deployed a hybrid search engine combining semantic search, vector embeddings (Elasticsearch ELSER + Dense Vectors), and BM25 retrieval with intelligent LLM-based query generation using Gemini models. The system processes 25,000+ federal and private grant opportunities with 91ms p50 latency and 99.9% uptime on GCP/Kubernetes. Beyond federal grants, I built a complete data ingestion pipeline that processes IRS Form 990-PF filings from 756,000+ private foundations—parsing XML tax returns to extract organizational metadata, grant histories, financials, and identify funding patterns and anomalies across the philanthropic landscape, then bulk-upserting this data into our funders database using cursor-based pagination and retry logic to handle large-scale operations reliably. I've also built conversational AI interfaces including the "Talk to NOFO" chatbot, enabling users to query complex funding documents through natural language. My work spans the full stack from FastAPI async endpoints to Kubernetes deployment configurations with sophisticated health probes and resource management, managing both real-time search (sub-100ms) and background enrichment workflows processing 50+ grants concurrently. This role combines my expertise in distributed systems, machine learning, and production engineering to solve real-world problems in the public sector funding space.

As a Graduate Teaching Assistant for CS5800 Algorithms at Northeastern University's Khoury College of Computer Sciences, I played a vital role in enhancing the learning experience of graduate students. Working alongside Prof. Virgil Pavlu and a team of graduate and PhD TAs, I contributed to various aspects of this advanced course. My responsibilities included grading assignments and exams with meticulous attention to detail, providing constructive feedback to foster student growth, and conducting weekly office hours for personalized support. I excelled in clarifying complex algorithmic concepts, from fundamental data structures to advanced topics like Dynamic Programming, Graph Algorithms, and NP-complete problems. This role deepened my expertise in algorithms, C++, and Python while honing my skills in problem-solving and technical communication. This experience as a TA was mutually beneficial, allowing me to reinforce my knowledge while helping others grasp complex computational concepts.

During my 6-month co-op at Veeco in San Jose, California, I immersed myself in the semiconductor industry, applying data science to real-world manufacturing challenges. I developed advanced machine learning models using TensorFlow and PyTorch to predict boron wafer resistance, achieving a remarkable 6% average error rate. To overcome limited data challenges, I implemented innovative techniques like data augmentation and group normalization, while experimenting with CNN architectures. I also enhanced data quality by improving Python scripts for systematic data extraction and structuring. A key project involved developing visualization code for channel data from manufacturing tools, enabling better pattern recognition and decision-making. Through interactions with industry experts, I gained deep domain knowledge in semiconductor manufacturing. My contributions extended to resolving data saving issues by revising Work Instructions, significantly improving data reliability. This experience honed my skills in process optimization, data mining, machine learning, and technical communication, while providing valuable insights into the semiconductor industry.

At NYU's Dynamical Systems Lab, I made significant contributions across various projects:

• Developed and crafted two successful grant proposals, securing $2.1 million in research funding.
• Spearheaded the development of machine learning models to predict ICU patient mortality rates, achieving 90% accuracy and contributing to improved patient care strategies in healthcare analytics.
• Leveraged the MIMIC IV dataset, containing over 300 million clinical observations, to develop and validate predictive models, demonstrating proficiency in handling and analyzing large-scale, complex healthcare data.
• Applied data analysis and information theory to study causal relationships between gun prevalence, mass shootings, and media, leading to a publication in Nature Human Behaviour.
• Created information theory-based models of zebrafish behavior, published in Flow: Applications of Fluid Mechanics.

This experience solidified my expertise in machine learning, causal inference, and data analysis, while honing my skills in research writing, mentorship, and interdisciplinary collaboration.

At Lost-Bytes, I bridged mechanical engineering with data science, developing an AI-enabled Digester prototype that converts food waste into organic fertilizer. This project introduced me to practical applications of machine learning in environmental solutions. I also created a data-driven food waste tracking software, analyzing nutritional components, temperature, and pH levels. This experience was pivotal in my transition to data science, showing me how data analysis can drive sustainability and efficiency in real-world systems.

My time at WindAid Institute was transformative. By manufacturing and installing wind turbines for off-grid households, I witnessed firsthand the impact of sustainable technology. This experience not only honed my engineering skills but also opened my eyes to the potential of data-driven solutions in addressing global challenges. It inspired me to explore how data science could be leveraged to optimize renewable energy systems and improve lives on a larger scale.

Key innovations included a 3D printed nose cone, improved plenum design, and the introduction of flax composites. Our efforts led to an impressive 8th place finish at Formula Student UK, including 6th in the grueling 22km endurance event.