Prudhvi Krishna Thandra: Advancing Data Analytics, Machine Learning and Intelligent Automation Through Practical Innovation
The growing integration of intelligence, data analytics and intelligent automation is transforming how organizations address complex challenges and improve operational efficiency. Professionals who combine technical expertise with a practical understanding of business requirements are increasingly valuable in this evolving landscape. With experience spanning data analysis, information governance, machine learning and software development, this technology professional has built a foundation focused on applying solutions to improve accuracy, productivity and decision-making. That foundation is backed by a body of research that ranges from medical image processing to cloud databases to financial forecasting.
Building a Strong Foundation in Computer Science and Data Analytics
A strong academic background has shaped this technical trajectory. After completing a Bachelor of Technology in Computer Science and Engineering at Kalasalingam Academy of Research and Education in India, Prudhvi pursued a Master’s degree in Computer Science at Wright State University in Dayton, Ohio, completing the program in April 2024. This progression gave him a foundation for exploring methods, analytical problem-solving and the practical application of emerging technologies.
His professional interests are reflected in his experience with Python, SQL, R, Tableau and Power BI, alongside machine learning frameworks and workflow automation platforms. These skills support an approach that connects data processing and visualization with the development of systems designed to improve organizational performance. Many of them also appear in his research, where Python libraries such as scikit-learn, NumPy, pandas and Matplotlib were central to his work on bot detection.
Early Projects, Early Results
The pattern was visible well before the degrees were finished. In March 2018, an automated Water Management System he built for crop irrigation was recognized among the Top 3000 Startups in India by IIM Calcutta Innovation Park and the Department of Science and Technology, Government of India. By monitoring and adjusting irrigation precisely, it achieved a 40% reduction in water usage and a 25% improvement in crop yield.
Around the same time, he took on the role of Vice Chair of the Association for Computing Machinery (ACM) Student Chapter, a position he held from August 2018 to May 2020, supporting collaboration and engagement in computer science. A brief stint as a web developer at Smartbridge Solutions in Hyderabad added a grounding in e-commerce, customer experience and conversion optimization, a reminder that technology only matters when it serves the people using it.
From Agriculture to Applied Machine Learning
That mix of curiosity and practicality led naturally to machine learning. During an internship at Avancer Software Solutions in Chennai from January to May 2020, he worked on
image-processing techniques and convolutional neural network algorithms for plant disease detection, achieving a reported 33% improvement in accuracy. The experience informed the collaborative publication Plant Leaf Disease Detection Using CNN Algorithm, published in 2021 in the International Journal of Information System Modeling and Design (volume 12, issue 1, pages 1-21).
Taking the Work into the Enterprise
By September 2020, he had moved into industry as an Assistant System Engineer at Tata Consultancy Services in Hyderabad, where he stayed until September 2022. There he contributed to data analytics, information governance and workflow optimization. His initiatives reduced data errors by 15%, improved data accuracy by 15%, and lowered annual data storage and processing costs by 5%. A data extraction tool he developed saved 15 hours of team effort weekly and improved analytical efficiency by 30%.
The machine learning interest followed him there. He developed a chatbot for USAA using the RASA framework, which reportedly reduced customer service response time by 50% and increased customer satisfaction ratings by 25%. Rather than treating automation as an isolated technical exercise, the work connected model-driven functionality with tangible service objectives. He also designed ServiceNow forms and automated data processing mechanisms, with reported gains of 10% and 20%, while his testing work contributed to a 30% improvement in product quality.
Returning to Research
Graduate study brought a more systematic phase. Prudhvi’s research papers share a habit of comparing methods head to head and being candid about what works and what does not.
His later research on cloud datastores echoes the enterprise focus of earlier years. He proposed using machine learning to partition data automatically in NoSQL systems built on HDFS, HBase and Apache Zookeeper, sparing developers the burden of manually analyzing access patterns.
The image-processing thread, first seen in plant leaves, resurfaced in medical imaging. In Advancements in Image Denoising: K-SVD, NLM, and PGPD Algorithms, he compared nine techniques on classic test images and on retinal OCT scans with unknown noise. K-SVD achieved the highest PSNR on the retinal images at 26.68 dB, ahead of BM3D at 25.73 dB, while PGPD struggled with salt-and-pepper noise, where the simple median filter won. Even the best denoising did not significantly improve retinal layer segmentation, a reminder that better-looking images do not always mean better results.
AlphaNAS: Efficient Architecture Search for Large Language Models looked ahead to the resource demands of modern AI. It proposes a theoretical framework combining Monte Carlo Tree Search with a surrogate model that predicts an architecture’s performance and computational cost, so promising designs can be found without fully training each one.
In Comparative Analysis of Machine Learning and Deep Learning Models for Real Estate Stock Price Prediction, he evaluated eight models on five years of daily closing prices for three Vietnamese real estate companies. The recurrent models (RNN, GRU, LSTM) and Random Forest were the most consistent performers, while ARIMAX, FFT and TimesNet lagged behind.
Finally, in Anomaly Detection in Facebook Ads Bidding, he used Kaggle data covering roughly 3 million bids from 1,984 bidders, 103 of them bots. Gradient boosting reached an AUC of 0.9169 after careful feature engineering, notably entropy-based features describing how widely a user’s devices and IP addresses were spread. It outperformed the Bayesian BIRDNEST method at about 0.84 and placed his approach in the top 15% of the leaderboard.
Where It Leads
Looking back across enterprise analytics, chatbots, medical imaging, forecasting, cloud systems and agricultural automation, a consistent theme emerges: using technology to solve defined problems and deliver measurable outcomes. His research reinforces this, testing ideas against evidence and reporting limitations honestly.
Thandra’s professional journey shows how a foundation in computer science, combined with applied projects and research, can support continued work in machine learning, analytics and technology-enabled innovation.
