Jan 2024 – Oct 2024 · 9 mos

Mar 2020 – Feb 2023 · 2 yrs 11 mos · India

• Built Ask.fi from scratch
• Developed NIDS and HIDS SIEM systems for epifi infrastructure

Mar 2018 – Dec 2019 · 1 yr 9 mos · Bengaluru, Karnataka, India

• Built ranking framework, enabling multiple signals to be used for ranking search results on dunzo app. Created popularity, merchant-score, user-score signals which led to 5% improvement in metric Search-to-Click ratio.
• Tech stack - golang, django admin, postgres
• Added a language model based query classifier to search, which classifies query to different categories. Added products indexing, and Phonetic search. The full system improved CTR from 22% to 56%
• Tech stack - elasticsearch, node js, aws kinesis - lambda
• Built Service for indexing and maintaining search data for dunzo. All the stores in dunzo are created (or updated) through this system, the service has features of detecting a possible duplicate store.
• Tech stack - django DRF, mongodb, elasticsearch, aws kinesis - lambda
• Enabled creating food orders from google maps to dunzo.
• Tech stack - golang, postgres

Jul 2015 – Feb 2018 · 2 yrs 7 mos · Greater Bengaluru Area

• Part of ​thematic​ ​team,​ ​which​ ​generates ​themes​ ​from​ ​​structured data,​ ​and​ ​auto-generate​s ​high​ ​click​ ​rate web ​pages.​ ​These​ pages​ ​capture​ ​various​ ​aspects​ ​such​ ​as​ ​seasonalilty of queries​ ​and​ ​popular​ ​products
• ● Facet,​ ​Sort,​ ​Pagination​ ​on​ ​Autogenerated Pages
• ○ Developed​ ​filters,​ ​sorting​ ​and​ ​pagination​ ​options​ ​on​ ​pages
• ○ This​ ​feature​ ​made​ ​the​ ​otherwise​ ​static​​ ​pages​ ​having​ ​richer​ ​control​ ​for​ ​users
• ○ Automated​ ​integration​ ​and​ ​testing​ ​this​ ​feature​ ​for​ ​any​ ​merchant
• ● Pipelines​ ​Refactoring​ ​&​ ​Automation​ ​:​ ​​Stabilized​ page​ ​generation​ ​system​ ​and​ ​​automated​ ​manual integration​ ​process​.​ ​Built​ ​tools​ ​to​ ​monitor​ ​visits,​ ​revenue​ ​of​​ ​pages
• ● Spike​ ​detection​ ​​:​ ​Developed​ ​alert​ ​system​ ​for​ ​spikes​ ​in​ ​ongoing​ ​data​ ​values​ ​using​ ​k-means​ ​clustering
• ● Theme-ranking​ ​with​ ​language​ ​model​ ​​(self-motivated):​ ​Compute​ ​probability​ ​of​ ​a​ ​theme​ ​being​ ​typed​ ​by​ ​user​ ​in​ ​a​ ​search engine​ ​and​ ​use​ ​this​ ​score​ ​to​ ​remove​ ​improbable​ ​theme​ ​names
• ● Automation​ ​of​ ​QA​ ​with​ ​ML (self-motivated)​ ​​:​ ​Used​ ​CNN ​to​ ​predict​ ​relation​ ​score​ ​of​ ​an​ ​image​ ​and​ ​text
• ● Automated​ ​generation​ ​of​ ​Relevance​ ​Report​ ​generation​ ​in​ ​BR-hackathon
• ● Tech​ ​stack​ ​:​​ ​Java,​ ​Python,​ ​Play,​ ​Scala,​ ​Cassandra,​ ​Solr,​ ​Qubole

Jul 2013 – Jun 2015 · 1 yr 11 mos

• Evaluation​ ​of​ ​Mobility​ ​characteristic​ ​in​ ​a​ ​social​ ​network
• ● In​ ​a​ ​simulated​ ​environment,​ ​analyzed​ ​Call​ ​Data​ ​Record​ ​(CDR)​ ​file,​ ​and​ ​deduced​ ​the​ ​mobility​ ​pattern​ ​of​ ​people. Considering​ ​this​ ​mobility​ ​pattern,​ ​derived​ ​a​ ​graph,​ ​in​ ​which​ ​edge​ ​shows​ ​connection​ ​among​ ​people
• ● Developed​ ​algorithm​ ​to​ ​find​ ​virality​ ​coefficient​ ​of​ ​a​ ​disease​ ​associated​ ​with​ ​individual