Arnab Sengupta

Arnab Sengupta

About

Detail

AI Growth Engineer at Noxx
West Bengal, India

Timeline


work
Job
school
Education
folder
Project
flag
Award
auto_stories
Publication

Résumé


Jobs verified_user 0% verified
  • Noxx
    AI Growth Engineer
    Noxx
    Nov 2025 - Current (8 months)
  • Noxx
    AI Agent Engineer
    Noxx
    Aug 2025 - Oct 2025 (3 months)
  • G
    AI/ML Domain Lead
    Google Developer Groups on Campus HIT
    Oct 2024 - Sep 2025 (1 year)
  • Alchemyst AI
    AI Engineer
    Alchemyst AI
    Oct 2024 - Jul 2025 (10 months)
Education verified_user 0% verified
  • Heritage Institute of Technology
    Bachelor of Technology - BTech, Computer Science and Engineering with Specialization in Artificial Intelligence and Mach
    Heritage Institute of Technology
    Jan 2023 - Feb 2026 (3 years 2 months)
Projects (professional or personal) verified_user 0% verified
  • R
    Rhyzm
    Aug 2023
    Rhyzm is a responsive music downloader web application that is built using python on top of Flask and Jinja framework. The music data is provided by Spotify API and the song is downloaded using Pytube python library with the help of Youtube Data API.
Awards verified_user 0% verified
  • I
    Ranked in 3rd Place at Mockup 3.0 Designathon
    IEEE CS MUJ
    Dec 2023
    Finished in 3rd place at the Mockup 3.0 Designathon organised by IEEE CS MUJ, as part of a duo.
  • I
    Ranked among the top 15 finalists in Hack It Out
    IIT Patna
    Oct 2023
    Ranked among the top 15 finalists in Hack It Out, a hackathon held by IIT Patna as part of its Techno-Management fest, Celesta 2023, where I led a team of four.
  • H
    3rd Runner Up at HackHeritage
    Heritage Institute of Technology Kolkata
    Sep 2023
    Attained the rank of 3rd Runner Up out of 90 participating teams at HackHeritage, an offline 24-hour hackathon held at Heritage Institute of Technology, where I led a competent team of six.
Publications verified_user 0% verified
  • e
    Is a Large Context Window all you need? Exploring Time To First Token (TTFT)-context size tradeoff for Autoregressive LL
    engrXiv
    Jun 2025
    Recent advancements in auto-regressive large language models (henceforth referred to as LLMs) have significantly expanded context window capacities, with Meta’s Llama 4 Scout achieving a 10 million token input length . This expansion is facilitated by techniques like Rotary Position Embedding (RoPE) and YaRN (Yet Another Rope extensioN), which encodes positional information through rotational transformations, enabling models to process longer sequences effectively. This advancement opens up a host of opportunities for the ubiquitious LLMs. Yet, attention mechanisms barely sub-quadratic in their nature. This means that extending context windows introduces challenges in latencies, especially in scenarios where even sub-second delays can resul
  • IEEE
    An Approach to Detect and Classify Potentially Suspicious Activity from Real-Time Log Data using Anomaly Detection Metho
    IEEE
    May 2024
    This paper aims to propose an intricate yet streamlined approach to incorporate Machine Learning into Cybersecurity by employing anomaly detection techniques to screen real-time log data and identify as well as classify suspicious activity. Our research focuses on avoiding misclassification with particular emphasis on minimising permissivity to ensure that threats with greater risk are not misclassified to a lesser priority. The paper explores a two-tiered strategy comprising an initial model dedicated to the identification of potentially suspicious log data through anomaly detection methods, serving as a preliminary filter. Subsequently, a secondary model is tasked with the responsibility of classifying the identified threats into distinct