Dheeraj P Anikar Email & Phone Number

Santa Clara, California, United States

Grader for EE559: Machine Learning.

United States

Involved in AI-Safety readings, discussions, and projects on the future of AI and techniques to improve AI safety.Projects:Metal Neural Analysis: Won the prize for improving a classifier performance to detect trojaned neural network. This was done by transforming the output features of the hidden layer in the network, by using Singular Value Decomposition to retain significant features based on the significance of their singular values and pass it onto the next hidden layer.Adversarial Attacks: Implemented FGSM and PGDM attacks on a well-trained Deep Neural Net, to confuse it's output predictions. This was done to showcase how a sufficiently well trained Neural Network is still susceptible to breaking down to adversarial attacks

Grader for EE503: Probability for Electrical Engineers and EE689: Computational Intelligence and Neural Learning

Berkeley, California, United States

• Utilized Amplitude Analytics and Python to analyze user behavior, retention rates, and subscription patterns within the app to assist business growth decisions.• Conducted an in-depth, critical review of the auto-scoring feature, uncovering its robustness across a wide spectrum of conditions and user settings. My findings contributed to a key role in refining and enhancing its accuracy.• Utilized LangChain (LLM application) framework to demonstrate automation of data analysis tasks on auto scoring analysis.• Skills: Amplitude Analytics, Data Science, Product Analytics, Pandas, NumPy, matplotlib, Scikit-learn, Requests (Rest API)

• Spearheaded Pin-Failure Mode and Effects Analysis (Pin-FMEA) to identify, quantify and prioritize potential failure modes and risks within manufacturing process to enhance quality control and to ensure optimal safety measures across BOSCH products.• Developed python scripts integrated with Pandas and NumPy for efficient data manipulation and utilized regular expressions for parsing and cleaning circuit netlists. Automated validation checks against pre-defined safety criteria through these scripts and realized an annual cost savings of 10,000 euros• Performed Design FMEA (D-FMEA) employing risk quantification methodologies, where I systematically assign Risk Priority Numbers (RPN) based on the severity, occurrence, and detectability of each identified failure mode, ensuring a prioritized and data-informed approach to risk mitigation• In a dynamic, cross-functional team environment, collaborated with HW/SW developers and QA specialists to offer data-based insights, facilitating the development of effective countermeasures against identified risks. These data driven strategies ensured all products surpassed/met the rigorous requirements of the ISO 26262 safety standard

Bengaluru, Karnataka, India

Tutored high school students and Class XII students in Mathematics and Physics

Bengaluru Area, India

Bengaluru, Karnataka, India

• Developed a deep learning model geared towards the identification of unattended luggage in CCTV footage• Personally captured and curated a diverse set of images to establish a comprehensive dataset, ensuring robust model training• Employed transfer learning techniques to capitalize on the existing architectures of MobileNet-SSD and VGGNet using Tensorflow• Fine-tuned the models by adjusting hyperparameters and layer configurations, ensuring optimal performance in real world scenarios

Master Of Science - Ms, Electrical And Computer Engineering (Machine Learning And Data Science)

Activities and Societies: AI Safety Club, Tennis Club Coursework: Fall 2022: 1. Linear Algebra for Engineering (EE-510) 2. Probability.

Bachelor Of Technology, Electrical, Electronics And Communications Engineering, 9.44 Cgpa

Relevant Coursework: Digital Signal Processing, Digital Image Processing, Machine Learning, Control Systems, Artificial Neural Networks.