Sudip Walter Thomas Email and Phone Number
Sudip Walter Thomas has been working at Amelia, which is a conversational AI company, as a Research & Development Engineer. Previously he has worked with ECOM Express Private Limited, which is a third party logistics & supply chain management company, as a Data Science Associate. He has completed his Master's degree in Mechanical Engineering from IIT Bombay, specialized in Manufacturing Engineering where he successfully completed a project on Automatically Grading Short Answers using Natural Language Processing. He is from the Land of Black Diamonds, Asansol, West Bengal. He has completed his B.Tech from Kalyani Government Engineering College, where he worked on a project called Energy Generation from Speed Breakers using basic bicycle's chain and sprocket mechanism.He likes playing Volleyball and has represented The Mechanical Engineering department in interdepartmental competitions winning the first prize. Thanks to his keen interest in music, he was a part of two of his college bands and was the convenor of Riyaz, the Music club of KGEC.
Useready
View- Website:
- useready.com
- Employees:
- 331
-
Machine Learning EngineerUseready Feb 2024 - PresentBengaluru, Karnataka, India -
Research And Development EngineerAmelia Oct 2022 - Oct 2023Bengaluru, Karnataka, IndiaRetrieval Augmented Answer Generation• Engineered a novel answer generation system to extract relevant informationfrom vast corpora & generate responses to user inquiries• Reduced around 99% cost per query, removing dependency on OpenAImodels, integrating Falcon-40B, LLama-2-70B via vLLM• Designed and implemented the pipeline for data acquisition, processing,& indexing through Lucene, optimizing retrieval efficiency• Leveraged embedding models to generate vector representations ofdocuments and queries, enabling semantic similarity-based retrievalZero-Shot and Few-Shot Intent Classification• Combined Setfit with LLMs gaining almost 25% intent prediction f1score, overcoming data scarcity, & made it available as a feature• Used Llama-2 after finetuning with LoRA, reducing GPU utilisation,time, resources, & leveraged gRPC calls to establish cross domain communicationfor efficient and fast information sharing• Conducted extensive research, testing& prompt engineering of variousLLMs & embedding models, deploying the most effective solution -
Associate Data ScientistEcom Express Private Limited Jun 2021 - Oct 2022Gurugram, Haryana, IndiaDelivery Centre Centroid• Worked closely with the ground team to recommend optimal deliverycentre geocoordinates using Kmeans clustering algorithm• Saved around 30% Last Mile operational cost using past data, recommendingoptimal dc location following business constraints• Managed relocation, split, closure cases of DCs on PAN India level• Visualized the results on a folium map, with heatmaps, hotspots etc.Box and Flyer Prediction• Trained a deep learning image classification model using VGG16 on Tensorflowto predict Box & Flyer images with 94% accuracy reducing morethan 14% cost per shipment• Automated the code using multithreading to download, load& process30 lakh images daily using boto3 from s3 bucket on ec2 instanceGeocoder Misrouting• Input Address -> Geocoder -> Address Coordinates• Processed and analysed geocoder data via sql queries, recommendingways to increase accuracy of geocoder, minimizing overall misrouting• Analyzed trends on a PowerBI dashboard to find patterns of intercityand intracity misrouting, distribution percentages etc• Converted address text to vectorized form using Tf-IDF and fed intoLSTM model to classify bad addresses with 96% accuracy -
Department Placement CoordinatorPlacement Office, Iit Bombay Jul 2020 - Aug 2021- Devised an efficient architecture of activities to train 70+ students on departmental level by ideating GDs, PIs, formulating tests, etc.- Teamed up with a group of 55+ DPCs to examine the resumes of 1600+ students across 23 departments for preserving authenticity- Acted as a bridge between the placement cell & students for smooth functioning of activities & flow of information among them -
Interview CoordinatorPlacement Office, Iit Bombay Dec 2019 - Jul 2020- Coordinated with a team of 250+ members for interviews of 1600+ students- Assisted in conducting Pre-placement Talks and Tests for 15+ firms. -
Convener At Riyaz, The Music Club Of KgecKalyani Government Engineering College Jul 2017 - May 2018• Coordinated with a team of 100+ members to design the architecture and management of cultural programs in Espektro'18, the Annual Techno-management cum Cultural fest of KGEC.• Organized and implemented various cultural activities all the year-round though Riyaz, in KGEC. -
Cultural Secretary, Rishi Bankim Chandra HallKalyani Government Engineering College Jun 2017 - May 2018• Spearheaded a team of 50+ members to successfully organize and supervise the events at Alfresco'18, the Annual Hall Day cum Reunion of Rishi Bankim Chandra Hall.• Responsible for the management and implementation of various cultural activities in RBC Hall.
Sudip Walter Thomas Education Details
-
Manufacturing Engineering -
Mechanical Engineering -
Asansol Collegiate SchoolScience
Frequently Asked Questions about Sudip Walter Thomas
What company does Sudip Walter Thomas work for?
Sudip Walter Thomas works for Useready
What is Sudip Walter Thomas's role at the current company?
Sudip Walter Thomas's current role is ML Engineer || LLM, ML, DS || Transformers, Tensorflow, PyTorch || M.Tech || IIT Bombay.
What schools did Sudip Walter Thomas attend?
Sudip Walter Thomas attended Indian Institute Of Technology, Bombay, Kalyani Government Engineering College, Asansol Collegiate School.
Who are Sudip Walter Thomas's colleagues?
Sudip Walter Thomas's colleagues are Rahul Jain, Jay Pandey, Anubhav S., Pooja Gudigar, Yash Gupta, Lithin Kumar, Sachin G..
Not the Sudip Walter Thomas you were looking for?
-
Sudip Walter Thomas
Bengaluru
Free Chrome Extension
Find emails, phones & company data instantly
Aero Online
Your AI prospecting assistant
Select data to include:
0 records × $0.02 per record
Download 750 million emails and 100 million phone numbers
Access emails and phone numbers of over 750 million business users. Instantly download verified profiles using 20+ filters, including location, job title, company, function, and industry.
Start your free trial