Building edge AI systems to assist surgeons and better understand the surgeries to allow for improved feedback and training.

Learning the latest in AI and MLOps, and building a network of entrepreneurs, founders and investors in US, Korea, India and Taiwan.

1. Marketing services and products for generating new business opportunities.a. Mobile Medical Units (customized as per requirement and need) b. FA6048 (Advance Fuel Treatment)c. Medical Consumables2. Identifying potential clients - contacting /visiting /follow up/maintaining all the communication records /giving presentations to them.3. Following up with existing clients - providing support & assistance to them.4. Contacting different vendors for medical & other equipment - getting best deals 5. Raising Purchase Orders /Quotations6. Updating higher management regarding potential clients& arrange meetings for the same.7. Projects Management - Visiting body building work shop for supervision of work on different stages.8. Preparing letters /tender documents / vendor registration.9. Attending exhibition and Seminars on behalf of the Company.10. Planning and executing promotions.

• Invented a patent-pending feature on our IoT device, which controls power to front-panel of the washer/dryer it is fitted in and allows resetting its control board. This eliminated false errors and dropped machine servicing time from hours to minutes. • Responsible for NPD & NPI of our IoT device that retrofits and enables smart control of commercial washers and dryers. • Manufactured 1000 v1 devices in collaboration with onshore contractors, while tactically ensuring supply of crucial chipsets. • Worked with select offshore partners for PCB design and manufacturing of v2 devices, their injection molded plastic enclosures, and assembly of custom cable adapters, in qty of 2000. • Finalized onshore and offshore manufacturing partners, developed a vendor base for chip components and other electronic parts, and established business accounts with Microchip and Raspberry Pi foundation for steady supply of their products. • Streamlined our mechanical manufacturing process, down to the sub-assemblies, to significantly reduce the time taken by 8X. • Developed design and packaging with onshore contractor, for our hardware kits, that would survive at least 10 round trips of transportation anywhere in the country.

Joined at Pre-Seed stage!Grew to Seed!

I worked on all aspects of product development lifecycle for 2 versions:Hardware Development• Led the process of customizing device tree structure, pinmux configuration, and bootloader and kernel modules of Nvidia Jetson Nano compute chip to adapt to its carrier board.• Identified optimal hardware components for night vision capability in our next generation device by conducting extensive assessment of different optical filters, infrared illuminators, and camera sensors and lenses for Nvidia Jetson Nano.System Integration and Testing• Integrated various sensors and network card, and stress tested our embedded platform to identify and rectify lockups of custom application processes, camera capture pipeline, and system throttling due to limited resources.• Developed intentional emission test modes for FCC certification of WiFi and Bluetooth Low Energy spurious emissions from the device, and test programs for assembly line approval of manufactured devices.Image Processing• Led the development of Triangulation and Trilateration algorithms for object position mapping from fisheye images to real space and vice-versa, and of Optical Refraction Model for correction due to Angle of Arrival of light rays in fisheye lenses.• Created novel background subtraction algorithm for thermal images to recover occluded objects of interest and to filter out unwanted objects being detected as false positives.Program Management• Led the process of software upgrade on our device to the latest host operating system, drivers, libraries and docker base image.• Managed up to date master BOM replete with device components, prices, lead times and feasible alternatives, to ensure manufacturing readiness.• Worked with contractor to produce 50 devices and managed site installation at NYP Queens.• Documented all experiments and procedures, and tracked updates to workflow processes, and presented them to project managers and CXOs in our weekly meetings.

1st hire, joined at Pre-Seed stage! Was full-time even before the co-founders, till we raised Seed!!

Managed 70 resident international students in helping them transition into US smoothly, so they can focus on their studies. Hosted monthly events to promote participation and interaction among all the residents.Partook in leadership workshops on global diversity and inclusion to create a harmonious living space among floor residents.

Worked as a Data Science Consultant for EmbedX Controls to help them in designing machine learning model pipeline in order to optimize energy consumption, making use of sensor data and energy usage patterns as input.

Worked as a student researcher in the VAP Lab at Columbia University, guided by Prof. Peter Belhumeur. Researched on improving the image recognition engine for the Leafsnap iOS application, and experimented with transfer learning by using pre-trained weights from DenseNet pre-trained on ImageNet dataset, to classify 55k leaf images into 185 classes. Best train/test prediction accuracy achieved was 98/72%. Then I trained a second model using DenseNet acrhitecture without any pre-trained weights and achieved a train/test prediction accuracy of 71/65%. Further, I also tried experimented training using the recent Multi-Attention CNN model. All these experiments involved extensive use of packages of Python, Tensorflow, Keras, Caffe, SkLearn, Docker, Jupyter, AWS & Linux.

Course Assistant for EECS-6720 Bayesian Models for Machine Learning. Instructed by Prof John Paisley.

Responsible for building predictive machine learning classifier models for measuring sentiment polarity, recognizing entities and checking topic presence in customer feedback, available as text data. These models were based on deep learning and machine learning techniques, including Recurrent Neural Networks, Convolutional Neural Networks and models based on random forest search. I created a novel encoding technique for converting text data to vectors to pass as memory efficient input into the machine learning models. All this was done by extensively using packages of Python, SciPy, Pandas, SkLearn, Tensorflow, Keras, NLTK, Gensim, Jupyter & AWS and while following Git version control and Agile methodology. This technique helped to build a viable sentiment classifier engine from text data available from half million customers feedback data. Finally, I also visualized and presented the results of these predictive models to senior management including the CIO of the company, which they used in their client presentations.

Analyzed past 5 years' equipment failure data, for the equipment used in the projects and built predictive models to learn about a new product's failure timeline/probability, using MatLab. This helped in analyzing the equipment's lifecycle and hence it's management as it was being used in substations. This analysis helped me in streamlining the replacement process of faulty substation automation & control products supplied for projects in Qatar to reduce the turnaround time between a warranty request and replacement. This led to improvements in quality control of Control & Relay Panels and Automation Systems supplied to electric substations in Qatar. Also, I was assigned project process flow responsibilities, wherein I conducted cost impact analysis and risk analysis for electric substation control and automation projects in India.

Responsible for finding out some strategic areas of business expansion for the company, wherein I assessed the LED lighting markets of India to identify potential partnership opportunities for a foreign OEM with Indian firms and compiled the recommendations. As part of that I also studied the Indian market for drones / Unmanned Aerial Vehicles and discussed with a few Indian firms for potential partnership with a foreign OEM.

Analyzed the 2013 Defense Procurement Procedures and reported the changes with respect to the older version of 2012 to help my managers easily gauge the differences. I also researched about the Indian Hotel Management industry and agreements and presented the findings to my managers, and this analysis helped them in identifying new opportunities and make new clients for business consulting.

Evaluated different stages involved in the production process of various electric cables manufactured in Paramount Communications Ltd factories located in the National Capital Region of India. Also, experienced first hand the quality assurance and testing procedures for electrical cables at those factories.

Conducted empirical experiments to test the fidelity of a current technology for calculating Snow Water Equivalent using acoustic signals, which involved designing an algorithm to measure the distance between the sound source and object of reflection used in the experimental setup with the help of Matlab. Then, I examined the results of the current method and identified the error between the distance obtained from the experiments and the actual distance and provided a root-cause analysis of the entire experimental process to diagnose and eliminate the error.

In this industrial training, I evaluated BEL’s product portfolio and field analyzed their various industrial processes for familiarization. Later, I was also trained on some of the PCB testing techniques.

In this defense research lab, I implemented the technique of Steganography using Matlab software and submitted a production .exe package that could be run standalone on a windows machine. Herein, I developed an algorithm to encrypt messages and embed them in images, without causing perceptible changes to the images, using the One Time Pad encryption technique.