×
0 followers0 following
Akashprabu Athikombai Chandrasekaran
California, United States0 followers0 following
- About
- Bachelor's
- Master's
- IELTS
Male
California, United States
Work ExperienceMLOps Intern
TheSmartBridge - Internship
May 5, 2020 - Jun 30, 2020 · 2 mos
Telangana, India
• Developed a config-driven end-to-end testing utility using Terraform, covering critical edge cases for cloud infrastructure.
• Identified and resolved design flaws in data models and infrastructure, enhancing system performance through creation of key metrics.
• Integrated an image segmentation model into a cloud-deployed product, using Triton and IBM Cloud to deliver the model as a service.
Software Development Engineer Intern
HyperWork - Internship
Aug 2, 2021 - Nov 30, 2021 · 4 mos
Lower Saxony, Germany
• Collaborated with the design team on responsive design, focusing API design, database migrations, and server profiling.
• Configured a performance and fault monitoring system, optimizing resource usage by 23% through redundant processes.
• Achieved 40% faster data retrieval by optimizing queries and the Python access layer, boosting efficiency and scalability.
Software Development Engineer Intern
CSG International - Internship
Dec 6, 2021 - Jul 10, 2022 · 7 mos
Karnataka, India
• Built a Go-based transformation tool for data migration from legacy data sources, reducing task execution time by 60%.
• Automated daily key-value retrieval from AWS Redis Cache using Bash scripts to enhance performance analysis.
• Implemented Redis cache for session retrieval and reducing latency by 20% for MVNOs like BillionConnect, Holafly.
• Configured a robust monitoring solutions using Prometheus and Grafana, creating dashboards, metrics and alarms.
• Converted traditional Docker files to Containerfiles for all microservices, optimizing containerization workflows.
• Troubleshot critical design flaws, production defects, and bugs, resolving them through analysis and business validation.
Software Development Engineer
CSG International - Full Time
Jul 20, 2022 - Jul 21, 2024 · 2 yrs
Karnataka, India
• Led end-to-end development of CSG’s core product, Ascendon Rating & Charging—a cloud-native SaaS for telecom billing and revenue management, enabling real-time charging with providers like Google Fi, T-Mobile, and AT&T.
• Designed 5G microservices to handle the policy charging and rating of 10 million users’ profiles with 5G-supported plans, which will process depletion in the subscriber’s available entitlement balance in real-time.
• Spearheaded the creation of a scalable microservice for data management, leveraging AWS services - ECS, S3, and Kinesis Firehose to stream call-detailed records, improving efficiency by 80% and reducing costs by $10,000 per quarter.
• Transitioned 4G and 5G microservices from AWS cloud-dependent solutions to a cloud-agnostic architecture, enabling deployment on Kubernetes/OpenShift across any cloud or on-premises platform, and eliminating code dependencies.
• Developed and deployed 20+ RESTful APIs in Go, reducing response time by 30% and enhancing concurrency.
• Leveraged Azure DevOps to streamline code development, and setup CI/CD pipelines for K8s and AWS deployment.
• Implemented Infrastructure as Code using Terraform to manage AWS resources and deploy Kubernetes configurations.
Software Development Engineer Intern
Amazon - Internship
May 19, 2025 - Aug 22, 2025 · 3 mos
Washington, United States
• Designed and implemented the ModelRegistry microservice using Java, API Gateway, AWS Lambda, and DynamoDB to onboard and manage 1000+ ML models across internal hosting platforms, reducing manual effort by 90%.
• Built 12+ OAuth2-secured RESTful APIs to manage model metadata, artifacts, QA results, and deployment states; adopted by 5+ internal teams to enable seamless integration and end-to-end automation.
• Enabled agentic model context delivery through Model Context Protocol (MCP) and built a Cloudscape portal (UI) to improve model traceability, deployment insights, and real-time monitoring for 50+ ML engineers and scientists.
• Contributed 80+ unit and integration tests, achieving 90% code coverage and supporting reliable CI/CD pipelines with minimal regressions.
ProjectsAI Powered Real Time Surveillance
Role - Developer
for Personal
Sep 1, 2020 - Nov 30, 2020
Regular Sanitization of workspaces is mandatory to maintain personal safety in co-working spaces. It is very difficult to maintain a log of sanitisation with personal supervision. In our project, we leverage the technology (IoT) to build a smart system to monitor sanitisation effectively at a low cost. The sanitisation log is taken with the help of hardware components like NFC Tags, and NFC-enabled Band or Mobile App. On the other hand, the log is stored and maintained in DynamoDB. A Face Mask Detection Platform uses a Deep Learning Technique to recognise if a user is not wearing a mask, which is implemented in CCTV cameras. If the camera captures an unrecognised face, a notification can be sent out to the administrator through SMS as well as email. In addition to that, a log is maintained, which is stored in the Firebase along with the face of the person who is not wearing the mask. All the information is retrieved from Firebase and DynamoDB to the actual website, where the admin can access, analyse, and report the activities. This system can be implemented in several workplaces.
Soldier Strap
Role - Developer
for Personal
Apr 1, 2021 - Jun 30, 2021
Soldiers can be assigned to stay in tough and dangerous places like deserts, the Himalayas, etc.. guarding the country. At these times, we need to know the status of the soldiers from time to time so that we can help those soldiers when they are facing health issues, etc... In order to overcome this problem, our project, Soldier Strap, is a device used to monitor and communicate the status of the soldier to the base station. The data that is transferred from the Soldier Strap to the base station contains information such as GPS Location, Body Temperature, Pulse Rate, and Blood Oxygen level of the soldier over a long distance without network connectivity.
WashBot
Role - Developer
for Personal
Jun 1, 2020 - Jun 30, 2020
Washbot is a hand-washing monitoring device that allows the user to maintain their hand hygiene and record data about it. The device’s functions are fully automated and hence can be maintained at low cost.
The user reaches the device and places his/her hands under the device. As the ultrasonic sensor is triggered, a sequential procedure of events begins. At first, the faucet is controlled, and water is dispensed for the user to rinse their hands. The hand-wash solution is dispensed, and the camera starts capturing the hand-wash procedure done by the user. The live feed is sent to the server from the device, where it is analysed using a machine learning model.
The model is designed under the hand-washing standards that are set by WHO, of which five basic steps are selected, and the data of those steps are used to train the model. This model is built on the VGG-19 model framework with 16 layers of CNNs and two fully connected layers, concluded with a softmax layer. The five basic hand-washing steps are labelled as classes for the model to predict the steps done by the user in the live feed. After analysing the live feed, the recognised steps are marked as done in the database, and the remaining steps that are neither recognised nor done by the user are marked as not done in the database. Once the user completes washing his/her hands, water is dispensed for the user to rinse their hands, and the record of hand-washing data will be displayed on the LCD screen.
This device is developed with the intent of maintaining hand hygiene and preventing contagious diseases in a public institution like a hospital, where interaction between people is very high. As for the hospital, the doctors, nurses and patient attenders are the disease transmitters as they interact with patients often. To prevent such a disaster, this device will be helpful, as it can wash their hands and the database assists the user in correcting the way they wash their hands.
Disease Classification in Paddy Crops
Role - Developer
for Personal
Jan 1, 2020 - Feb 29, 2020
Early disease detection plays a major role in the protection of paddy crops. In earlier days, the detection of disease was done through seeing or by examining in a laboratory. The observation made visually needs experts, and it might vary for each individual, which leads to error and laboratory testing requires more time and might not be able to deliver the outcome within a time. To get the better of this issue, image processing-based Machine learning approaches are implemented to detect the diseases and classify those diseases. We mainly focused on rice(Oryza sativa) diseases. The images contain the leaves and stems, which are affected by disease, collected from the paddy fields. Totally, we have collected 2000 affected paddy plants from Kaggle and rice fields. The dataset contains five different classes of diseases: (1) Rice Blast, (2) Bacterial Leaf Blight, (3)Sheath Blight, and (4)Healthy leaves. The early detection of diseases will help farmers to increase their yield.
Personal Electrical Women Safety Device
Role - Developer
for Personal
May 1, 2021 - Jun 7, 2021
Personal Electrical Women Safety Device for women is designed with the aim of building a device with integrated functionalities for the safety of women. The keyword from the person, which is a voice-based command, is received through the microphone, and the voice is recognised by the mobile application-based voice assistance. This application then stores the voice recording / Location of the person in the cloud, and it can be accessed from a Web portal. This application sends an SOS alert message carrying the GPS location to the emergency contacts. This application also sends the message from the sender controller to the receiver controller. This device can be used even in places where there is no network connection. The data sent from the device covers a long range. Even if it exceeds the specified range, it can form a mesh network with the other devices of the same type to expand the range in order to transfer the data containing the GPS location. This device works with or without the mobile application. When the device is connected to the mobile application, we can use voice recognition to send data through three-way communication (LoRa/ SMS / Web portal). Without the mobile application, we can send data through LoRa alone by pressing the push button on the device. The data sent through LoRa contains the Location, Pulse Rate, and Temperature of the person, either through a mesh network or Direct communication. The result of this design will allow women to feel safe and secure during the incident.
Patent Application No: 202141025203
Research PapersEnhancing Heart Disease Prediction Through KBEST-PCA Fusion Feature Selection and Ensemble Modeling With Gaussian Naive Bayes Boosting
Jul 19, 2023 - International Journal for Multidisciplinary Research (IJFMR)
Heart disease is a prevalent health condition with significant implications for patient health and well-being. Accurate and timely diagnosis plays a crucial role in effective treatment and management. In this study, we propose a combined approach using SelectKBest, Gaussian Naive Bayes (GNB), and Gradient Boosting Machines (GBM) to develop a robust predictive model for heart disease diagnosis. The SelectKBest algorithm is employed to identify the most informative features from the Statlog Heart Disease dataset. Statistical measures such as the chi-squared test are utilised to select the top K features that exhibit the strongest associations with the target variable. The selected features are then used to train a GNB classifier, capturing the probabilistic relationships between the features and the diagnosis of heart disease. Predictions generated from the GNB model are combined with the original features, creating an extended feature matrix. Subsequently, a GBM ensemble model is trained on the extended feature matrix, leveraging the sequential combination of weak learners to improve the overall predictive performance. To evaluate the effectiveness of the proposed approach, extensive experiments are conducted on the Statlog Heart Disease dataset. Performance metrics, including accuracy, precision, recall, and F1 score, are used to compare the combined SelectKBest-GNB-GBM approach against individual classifiers and existing methods.
PDF
Enhancing_Heart_Disease_Prediction_Through_KBEST-PCA_Fusion_Feature_Select_krafIve.pdf
VolunteeringVolunteer
at Way for life
Aug 1, 2023 - Aug 2, 2023
Environment
Collaborated with a youth-driven NGO, Way For Life, in creating 53 solar lamps, which replace traditional Kerosene lamps and help in creating a cleaner and brighter environment. With this great opportunity, we have helped the local communities in gaining access to clean, renewable energy and raising their quality of life in a sustainable fashion.