- Developed, implemented, and deployed to production several text classification models resulting in improvements in precision up to 300% compared to benchmark models.- Performed ongoing production model performance and iterative improvement.

- Developed a core logic for a customer support ticket model classification using a zero-shot learning model. Collaborated with a team of developers to create an application and deploy it to production.- Created datasets for machine learning model development using internal and external data scraped from various APIs.- Applied unsupervised machine learning techniques (SentenceBERT clustering) to create initially labeled datasets for model training.- Analyzed and visualized text and time series data to derive valuable insights with the goal to inform business decisions and provide guidance to clients.

- Development of an ElasticSearch search engine for document storage and retrieval for an intelligent open source virtual assistant (Qary).- Built a dockerized Django application prototype for the virtual assistant.- Working with a 100% remote team- Gitlab repository: https://gitlab.com/tangibleai/django-qary

Work with a 100% remote team building a smart automated material property prediction engine.Developed document database (MongoDB) for dynamic aggregation and storage of data on various stages of the data processing pipeline. Created a dockerized WSGI Flask demo application to showcase the material property predictor and feature finder tool to potential clients and investors.

- Initiated and managed research and development of low-cost, thin-film flexible solar cells.- Collected, analyzed and managed data from various sources using Python’s Pandas, StatsModels, Matplotlib, Scipy and Pymongo libraries.- Developed and managed MongoDB database for heterogeneous data aggregation and centralized access by other team members- Analysis of state-of-the-art literature, idea generation, coordination of a multidisciplinary team of scientists and engineers.- Developed new nanofabrication procedures, nanoparticle synthsis methods, and material-level and device-level testing methodologies.- Ensured safe work environment and developed OSHA-compliant procedures.- Coordinated collaboration between cross-functional internal and external teams.- Documented and maintained workflow procedures, reports, memos and manuals.

Selected projects: - Twitter Data Mining Sentiment Analysis: Collected Twitter posts of 2020 presidential candidates and financial influencers to observe any correlation between the overall tweet sentiment and the stock market movements. - Forecasting of Wait Times on Street Intersections: Predicted traffic wait times on busy intersections using a dataset provided by BigQuery-Geotab on Kaggle. Machine learning methodologies include linear regression, random forests, gradient boosted decision trees, Bayesian optimization, cross-validation and parameter grid search.- Exoplanet Classification: Applied k-means clustering, support vector machines, random forests and deep learning to classify space objects detected by NASA’s Kepler space telescope and identify exoplanets. Classification performed using support vector machines, random forests and convolutional neural networks.- Biodiversity App: Built full stack Flask application for exploration and visualization of bacterial populations in human navels. The interactive visualization dashboard is built using Javascript and Bootstrap/HTML. The app is deployed to Heroku server.- Mission To Mars App: Built full stack Flask application which uses Splinter and Beautiful Soup libraries to scrape the latest news related to Mars from NASA, USGS, Twitter and other websites on user’s request- Full Stack Visualization App - Google Store: Built full stack Flask application for interactive visualization of trends in Google Store applications. - ETL Project: Collected data and explored correlation between socio-economic statistical data from Census API and public health data, provided by Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Division of Population Health for 500 US cities. Transformed the data in Pandas and loaded it to MySQL database using the Python SQLAlchemy library. Executed SQL inner join to merge the datasets.

Ph.D. dissertation: Nanoscale active photonics and it's integration with silicon photonics.- Led research on extreme miniaturization of chip scale III-V semiconductor light emitting componentsand their monolithic integration with silicon photonics.- Fabricated nanoscale semiconductor lasers in a class 1000 cleanroom (Nano3 of UCSD) using a variety of lithographic, etching and thin (metal and dielectric) film deposition processes. Developed plasma assisted wafer bonding technology for integration of dissimilar materials.- Performed material and structural characterization using SEM/EDS,TEM, AFM, FIB, profilometry, reflectometry, and ellipsometry.- Developed analytical electromagnetic and thermal models of nanoscale optical cavities and subwavelength cross-section waveguides via finite element method (FEM, COMSOL). Performed coupled mode analysis of hybrid IIIV/Si distributed feedback lasers in Matlab and material electronic bandgap analysis in Silvaco ATLAS.- Designed and built two optical test stations using free space and fiber optics components. Modeled optimal beam profile of the probe laser via matrix formalism for Gaussian beam in Matlab.- Authored and co-authored publications in leading peer reviewed optics and photonics journals, gave multiple oral and poster conference presentations.- Designed IIIV/Si coupler for integrated photonic cirquits and fabricated the first prototype. Performed electro-magnetic analysis of the coupler with finite element time domain method (COMSOL). Developed the reactive ion assisted milling method to create 3D structures.- Developed the original robust nanofabrication process for optically pumped subwavelength metalo-dielectric laser in collaboration with other project team members.- Fabricated etch-free low-loss silicon waveguides. Developed the thermal oxidation process to achieve the geometrical precision of the waveguide structures in collaboration with other project team members.

Studied molecular beam epitaxy (MBE) of dilute nitrides. Performed Varian Gen II MBE system maintenance. Created graphical user interface for custom-made shutter control system. Fabricated the first prototype of room temperature sub-wavelength laser in collaboration with another research group.

Integrated process modules for optical proximity communication packaging. Performed process logging, progress reporting, literature search and analysis.

Epitaxial growth and investigation of N-containing III-V materials and optoelectronic heterostructures via molecular beam epitaxy.