def __init__(self):
self.name = "Mayank Prasad",
self.occupation = "Data Scientist",
self.likes = ["Travel", "Home Automation"]
About Me
Data Scientist with over 9 years of experience ensuring data correctness, deploying complex machine learning models and statistical modelling algorithm for identifying patterns and extracting valuable insights with domain understanding. Demonstrated ability to effectively engage with technical professionals and end users to identify and translate business requirements. Proven ability to manage group of talented professionals to develop valuable process solutions that satisfy business objectives. AI Enthusiast.
Portfolio
One of my significant achievements includes creating a data analysis and Machine Learning algorithm to determine the age of battery cells from various suppliers. The application forecasts battery aging and degradation ten years ahead of time using measurement data. This experience reflects my dedication to delivering solutions that Achieved enhanced and accelerated cell selection, thus saving significant cost and time of the diagnosis team
Beyond my core projects, I travel, involve in home automation, Finance and investments, and also contribute to CSR initiatives, firmly believing in the connecting with people, power of knowledge sharing and collaborative development. It's my way of staying engaged with various teams and with the latest industry trends.
Timeline
BMW
Mercedes Benz RD India
The National Institute of Engineering, Mysore
Delhi Public School, Bokaro
Dec 2024 - Present
Aug 2017 - Dec 2024
Aug 2013 - Jun 2017
Apr 1999 - Mar 2013
My Work
Cell Ageing Analysis.
Thermal Model Analysis.
Developed an XGBoost-based system to predict battery cell aging across 15 vendors and ~300 cell variants, using gradient optimization and statistical modelling to forecast degradation up to 10 years ahead under varying climatic, range, and performance conditions.
Applied matrix-based feature engineering and Bayesian classification on cell measurement data to sort battery cells into Mercedes-Benz AMG performance vehicle categories.
Streamlined cell selection for the diagnosis team, reducing cost and evaluation time via data-driven hypothesis testing.
Built a battery thermal management optimization pipeline for EVs by combining field data with Gradient Boosting (classification) and regression modelling for thermal cooling processes, deployed on AWS SageMaker.
Reduced model creation time from 720 hours to under 5 hours using supervised and unsupervised learning, minimizing manual intervention and human error in model generation.
Verified model outputs by running Simulink simulations in MATLAB against real-world field measurements.
Creation of driving profile and power profile for battery variants to ensure battery robustness.
Designed power profiles for lifetime simulation of battery capacity used as input for robustness.
Automated end to end application development and deployment to multiple departments using CICD
Emission Profiling.
Ai Assistant for ECU Network Communication
Modified BEVFormer to evaluate 3D object detection robustness under camera extrinsic de-calibration (simulating sidemirror displacement from pedestrian/cyclist impact) on the nuScenes dataset.
Demonstrated that even ±5° de-calibration of a single camera reduced mAP by ~28% (0.264 → 0.19), with multi-camera perturbations causing up to 73% degradation — highlighting the need for online recalibration in ADAS systems.
Camera De-calibration Impact on 3D Object Detection
Built an LLM-based assistant that converts Jira change requests into valid SyML specification code for BMW's vehicle E/E architecture, replacing a manual and error-prone workflow for network engineers.
Implemented a RAG pipeline with vector DB retrieval of past gold-standard changes to guide code generation, followed by automated grammar validation using Bazel.
Automated the complete flow from natural language input to Git commit and PR creation, significantly reducing turnaround time and domain expertise needed for spec changes.
Skills
Matplotlib
MATLAB/Simulink
Scikit-Learn
HTML/CSS
TensorFlow
Numpy
SQL
Seaborn
Python
FastAPI
Pandas
AWS SageMaker
AWS / Azure
Power BI
I had the experience of working with Mayank Prasad at the start of his professional career. He is very enthusiastic and has the zeal to learn new technologies and skills. On the technical front, he has the skillset of an advanced developer in Python, Matlab and Requirement Engineering. He also understands the importance of process in a team and has taken significant responsibility in this field. Few of his skills which are quite noticeable as a team member would be organized, proactive, putting forth his opinion and also taking criticism positively. I am sure, this will help him grow as a professional.
Indu Kumari (Senior Test Engineer at Munich Electrification (Battery Management System | ADAS)
Testimonials
Working with Mayank for the last 7 years gave me insights into his great problem solving capabilities and creativity. Mayank consistently proved his talent for thinking outside the box and bringing fresh solutions to challenging issues. It was a pleasure having Mayank on our project.
Guillaume Tyson (Ingénieur méthodes chez Daimler AG)
Mayank is good at Python based software development. He is well organized, hard working, fast learner and have good communication skills. All these attributes were critical when trying to work with different people under strict deadlines on core projects. His agility and adaptability are important skills which helped him in managing several situations and tasks of projects, specially during on-site trips. It was a pleasure to work with him for almost 3 years on several different software projects. Good luck.
Vishal Sharma (Simulation and modeling expert, Battery Modeling, Full Vehicle modelling and Simulation, EV, ex-Volvo, ex-Mercedes Benz, IIT Kanpur, Machine learning enthusiast)
Wanna talk / Hire?
Contact me with any questions or just to say
"Hi ..."
Mayank Prasad.
mayankpd@outlook.com
+91 9591079144