Building a generative AI-powered chatbot using RAG and fine-tuning open source LLMs, to enhance companies’ competitive edge.

Investigating the application of AI, machine learning, and statistics to democratize healthcare, by improving performance of sensors for medical diagnostic testing.• Devised deep learning-based approach to reduce sensor response time by > 5x, using ensembles of LSTM neural networks (Python, TensorFlow) for time series forecasting, uncertainty estimation, and transfer learning with a large-scale simulated dataset, enabling rapid testing of harmful molecules.• Designed a new biosensing paradigm by applying data visualization, classification and pattern recognition algorithms (Python, SciKit-Learn) to sensor array data, a step towards unprecedented robust, scalable, and low-cost sensors.• Invented algorithm using Morlet wavelet filtering and Fourier analysis (Matlab) which improved detection limits of thin film sensors by 10x, and released user-centric open-source app.• Built software (Python) and hardware to automate data collection, improving accuracy by 48% and increasing experimental throughput by 100x, enabling larger datasets for more generalizable models.

Led team of undergraduate and graduate students working on projects I curated. The 5 undergraduate mentees over 4 years went on to be co-authors on publications, presenters at national conferences, and embark on PhD degrees of their own.

Instructed Digital Systems undergraduate course focused on python and electronics, creating 30% of lab content.

• Developed production-ready firmware (C) and hardware, collaborating with global cross-functional teams, driving the companies push towards IoT and expansion into a previously untapped market.• Solved design flaws in products by troubleshooting customer issues under pressure and finding the root cause (temperature drift, static electricity), rescuing orders ($70,000+).• Designed new test procedures, using software (Python) and hardware to raise production yields by 5%.

• Engineered pipeline defect testing solution using eddy currents and data analysis (Python) with the potential to reduce operating costs by >20% and communicated findings to company stakeholders, in collaboration with GE.• Modelled physics of sending a rocket to the moon (Python), adding novel functionality.• Probed molecular behaviour of surfactants using dual polarization interferometry, providing valuable insights Procter and Gamble product development, presented to P&G stakeholders.

• Formulated and performed experiments to test software and hardware of a gas detecting camera and designed electronics in intelligent junction box, collaborated with a team in a partner company abroad.

• Quantified vibration in cryogen-free superconducting magnet system with laser Doppler measurements.