Harshit Manek
ML EngineerAI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Computational Research Scientist at Keck School of Medicine, USC | Single-cell & multi-omics | clinical/translational analytics | ML/LLMs
I’m a computational biologist/bioinformatics scientist passionate about using single-cell biology, multi-omics, and machine learning to accelerate translational research—especially in immuno-oncology. My background spans scRNA-seq + TCR repertoire analysis, scalable HPC pipeline engineering, and ML/LLM systems that make messy biomedical and clinical data usable (from controlled-access omics repositories to clinical/regulatory data standardization like CDISC SDTM). I enjoy work that sits at the intersection of rigorous science and real-world impact: building reusable analysis infrastructure, collaborating closely with biologists/clinicians, and translating complex results into decisions teams can act on. I’m always happy to connect with peers, collaborators, and teams working to apply science and technology to solve problems in health and biology.
University of Southern California
Master's degree, Computer Science (Artificial Intelligence)
August 1, 2021 – May 1, 2023
Cluster Innovation Centre, University of Delhi
Bachelor of Technology (B.Tech), Mathematics and Computer Science
January 1, 2013 – January 1, 2017
Apeejay Stya Education (Svran Foundation)
Mathematics and Computer Science
January 1, 1999 – January 1, 2013
Keck School of Medicine of the University of Southern California
Computational Research Scientist
July 1, 2023 – Present
Los Angeles, California, United States
Genentech
Data Scientist Co-op
May 1, 2022 – June 1, 2025
South San Francisco, California, United States
USC Information Sciences Institute
Graduate Student Research Assistant
March 1, 2022 – June 1, 2025
Elucidata
Data Scientist
February 1, 2019 – July 1, 2021
Delhi, India
IndiQus Technologies
Data Scientist
March 1, 2017 – February 1, 2019
IndiQus Technologies
Data Science and ML intern
March 1, 2017 – July 1, 2017
PaperKite
Data Science and ML intern at PaperKite
February 1, 2017 – April 1, 2017
New Delhi Area, India
Defence Research and Development Organization
Machine Learning Research Intern
June 1, 2014 – December 1, 2015
New Delhi Area, India
Cryptography, Application of Deep Belief Networks & Fuzzy Sets in Speaker Recognition | SAG Lab, Defence Research & Development Organization, Delhi
June 1, 2015 – December 1, 2015
• Explored various variations to RC4 stream cipher, like the Spritz algorithm. • Worked on Chaotic cryptography techniques, exploring 1D and 2D chaotic maps. Created hybrid maps and tested their efficiency for multimedia encryption. The encryption scheme yielded a UACI score of about 32%, and an NPCR score of about 98%. • Thoroughly explored Neural Networks on various applications like the MNIST dataset & Iris Dataset. Incorporated Fuzzy Neutrosophic Decision Making into Neural Networks. • Working on celebrity speaker recognition using Deep Belief Networks, Fuzzy Neutrosophic Decision Making, and Swarm Based Techniques like PSO.
Celebrity Recognition in Video using MAD and LDS | Semester Project, Prof. Shobha Bagai
March 1, 2015 – July 1, 2015
• In this project an attempt was made to combine two semi-supervised algorithms “Modified Adsorption” and “Low Density separation” to make a better algorithm. The graph based similarity measures of Modified Adsorption were combined with the Transductive Support Vector Machine of Low Density Separation. • Both The Algorithms were implemented in C using GNU Scientific Library and Armadillo. The Face Detection and SIFT feature extraction was done using OpenCV. ‘Scikit-learn’ python library was used to provide initial labels to faces and reject noisy images using IMFB dataset. • The rho-path distance and Multidimensional scaling algorithms were also implemented and were used on initial graph matrix to increase accuracy. • The results were obtained using Cropped Yale and IMFDB (Indian Movie Face Database) and the accuracies obtained were 85% and 65% respectively which outperformed both MAD and LDS taken individually.
Matrix based Cryptographic Schemes –Internship at SAG Lab, DRDO
June 1, 2014 – Present
An internship at the end of the first year of B.Tech, I worked under the guidance of Dr. S.K. Pal, on the project Matrix Based Cryptographic Schemes. Initially I started off by analyzing and implementing optimized algorithms for a various matrix based operations. In this project, I studied and implemented various cryptographic schemes in C, Python and MATLAB and also tried attacks on the schemes, and for some schemes the security was finally breached. The schemes that were cracked (whose security was compromised) were also ameliorated by borrowing some techniques and functions from the well-developed ones. Since the schemes required very high computation, to enhance the computation of the programs GMP library was used for the same purpose and for high precision in C. Prime numbers which play an important role were handled by the Rabin-Miller Algorithm. Cryptanalysis were carried out for many schemes and loop holes were identified in each one of them. Different linear algebra concepts were explored and applied to the Information Security domain. Some of the techniques used in the schemes were MPF (matrix power function), Abelian Groups, Cayley Purser algorithm, Wang and Hu scheme, Pseudo Inverse of matrices and Singular Value Decomposition. Overall I had an enriching experience in both the Symmetric and Asymmetric Public key cryptography.
Robotic Simulation Using Player/Stage Simulation Software | Dr. Manoj Agrawal, Associate Professor
June 1, 2014 – September 1, 2014
• Included simulation of various tasks like Box Pushing, Obstacle Avoidance, Path Finding, on Player/Stage Simulation Software • Worked on developing algorithms for surveillance of battle fields, and field/room exploration, and testing the same through simulations on Player/Stage; involved coordination between various systems running different bots. • Worked on developing algorithms for constrained task division and scheduling problem, for supporting a distributed environment.
3D chess
February 1, 2014 – Present
Developed 3D chess using jmonkey game engine.Added features of artificial intelligence to provide better game play of user with CPU.
Cultural Fit Analysis
The candidate's background includes significant academic research (DRDO, USC) and industry experience in data science and computational research, often in interdisciplinary fields like bioinformatics. This suggests a fit for organizations that value continuous learning, research-driven innovation, and cross-functional collaboration. The diversity of projects (cryptography, AI, robotics, bioinformatics) indicates a broad interest in applying ML to various domains, which could be a strong cultural fit for dynamic and evolving ML teams. However, the focus on highly specialized research might require assessment of their ability to adapt to more product-driven, iterative development cycles if the target role emphasizes that aspect.
Soft Skills & Operational Fit
The candidate's project descriptions suggest a strong problem-solving aptitude and an ability to work on complex, research-oriented tasks. The diverse range of projects, from cryptography to robotics and AI, indicates adaptability and a broad technical curiosity. The experience in both academic research and industry roles (Data Scientist at Elucidata, IndiQus, Genentech) suggests an ability to bridge theoretical knowledge with practical application. However, without specific behavioral assessment data, it's difficult to fully assess collaboration, stress handling, or communication clarity in a team setting.