Jun 2023 – Present · 2 yrs 9 mos · Lugano, Ticino, Switzerland · Remote

Aug 2022 – Jul 2023 · 11 mos · Bangalore Urban, Karnataka, India

• Created ML pipelines platform in the company using Kubeflow,
• handled maintaining, improving UX and setting processes for pipelines contributed by different teams.
• Automated training, QA and deployment of client specific ML models.
• Helped reducing engineering bandwidth and cost highly across company.
• Formed culture and platform for A/B experimentation and analysis at scale
• for various product usecases across different teams.
• Owned several internal tools - skit-pipelines, skit-calls, skit-labels.
• Integrated full CI/CD for various repos while reducing tech debt.

Aug 2021 – Oct 2022 · 1 yr 2 mos · Bangalore Urban, Karnataka, India

• Maintained multiple Spoken language understanding (SLU) models and fine-tuned ASRs on different clients data, both following best ml-lifecycle practices which acts as core components in our voicebots used for Airtel-dth, Dish-tv, Videocon-d2h, Titan, Meesho in their customer care experience automation.
• Contributed on automated model prediction calibration for SLUs using temperature scaling.
• Boosted the performance of our SLU API service by 20% with other big changes.
• Added full CI/CD for various internal tools.
• Created a process for acoustic-oos (background noise in audios specifically) classification which gives 91%+ F1 score on acoustic-oos when using our in-house ASRs in live production calls for transcriptions.

Jun 2021 – Oct 2021 · 4 mos · Atlanta, Georgia, United States

• 1500+ chatbots having unique personalities, can talk on anything, sometimes even better than a human + remember things.
• User can pick/give personalities and our system brings it to life.
• Worked on it while having 2100+ active users.

Apr 2021 – Jul 2021 · 3 mos · Bangalore Urban, Karnataka, India

• Built and deployed Brand Normalization API in production which reduced scraped noisy text -> brands in unsupervised fashion, achieved compression 94%, improved search quality.
• Validated data using Wikidata Knowledge Graphs and SPARQL, reduced bottleneck 2.4 hours -> 8 mins for 330,000+ Product titles.
• Managed, preprocessed data and built Fashion Text Attribute Extraction pipeline.

Jun 2020 – Aug 2020 · 2 mos · Mumbai, Maharashtra, India

• Built and deployed a system to detect people from videos like CCTV footage and extract faces, detect gender and age from it.
• Scraped and visualized regional news articles + Topic Modeling to analyze how words trend have changed in pandemic.

Oct 2019 – Aug 2022 · 2 yrs 10 mos · Bhubaneshwar, Odisha, India

• Machine Learning and Deep Learning Specialist in ML Team
• Started “How I Made It” a podcast by DSCKIIT, we talk to people who made it big in the Tech Industry.
• Started a weekly AI Research Newsletter to encourage more students in STEM research, with 500+ subscribers.
• Organizing tech sessions, built impactful projects, wrote technical blogs and increased 1000+ students engagement.