Amogh Targe

Amogh Targe

+•• ••••• •••••••••••@•••••.com

Specialist Programmer - L2 · Infosys · 3 yrs

  • Open to Bangalore · Pune · Noida · Gurgaon · Singapore · Dubai · London · Amsterdam · Berlin · New York · Tokyo · San Francisco · Mumbai · Hyderabad
  • Hybrid, Remote, On-site
  • In 30 days

Ask about Amogh

Skills

Claimed

Data ModerationGreenfield DevelopmentLLM Infrastructure DevelopmentAI-Powered Tool DevelopmentSmall Language Model ConceptsHuman Feedback IntegrationRLHF Workflow DesignSQL GenerationState Representation DesignReinforcement Learning Environment CreationMulti-turn Deviation Strategy DesignNVIDIA Triton Inference Server Usage

Proof of Experience

Infosys

Specialist Programmer - L2

Infosys · Jul 2024 – Present

  • 𝐀𝐩𝐩𝐥𝐞 𝐂𝐥𝐢𝐞𝐧𝐭
  • 𝐄𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐞𝐝 𝐃𝐚𝐭𝐚 𝐒𝐞𝐜𝐮𝐫𝐢𝐭𝐲 𝐩𝐥𝐚𝐭𝐟𝐨𝐫𝐦 𝐟𝐨𝐫 𝐏𝐈𝐈/𝐏𝐇𝐈/𝐏𝐂𝐈 𝐫𝐞𝐝𝐚𝐜𝐭𝐢𝐨𝐧, 𝐥𝐞𝐝 𝐚 𝐆𝐞𝐧𝐀𝐈 𝐫𝐞𝐜𝐞𝐢𝐩𝐭 𝐞𝐱𝐭𝐫𝐚𝐜𝐭𝐢𝐨𝐧 𝐚𝐩𝐩𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐚𝐝𝐨𝐩𝐭𝐞𝐝 𝐚𝐜𝐫𝐨𝐬𝐬 5+ 𝐭𝐞𝐚𝐦𝐬, 𝐚𝐧𝐝 𝐛𝐮𝐢𝐥𝐭 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐀𝐈 𝐬𝐲𝐬𝐭𝐞𝐦𝐬 𝐟𝐨𝐫 𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐑𝐮𝐥𝐞𝐬 𝐄𝐱𝐭𝐫𝐚𝐜𝐭𝐢𝐨𝐧 (𝐁𝐑𝐄) 𝐚𝐧𝐝 𝐥𝐞𝐠𝐚𝐥 𝐫𝐢𝐬𝐤 𝐚𝐬𝐬𝐞𝐬𝐬𝐦𝐞𝐧𝐭.
  • • 𝐃𝐚𝐭𝐚 𝐒𝐞𝐜𝐮𝐫𝐢𝐭𝐲: Engineered secure PII/PHI/PCI data processing using regex and ML algorithms deployed via Triton inference servers.
  • Architected a decoupled entity validation framework using the Strategy design pattern, scaling to new locales and reducing false positives by over 40% (e.g., utilizing Verhoeff and Luhn checks).
  • • 𝐆𝐞𝐧𝐀𝐈 𝐫𝐞𝐜𝐞𝐢𝐩𝐭 𝐞𝐱𝐭𝐫𝐚𝐜𝐭𝐢𝐨𝐧: Led production LLM adoption for an internal ticketing tool, resolving complex challenges like overlaid text and inconsistent orientation.
  • Automated Apple's policy adherence reviews by utilizing LLMs to generate a custom 13-point compliance score.
  • • 𝐋𝐋𝐌 𝐁𝐞𝐧𝐜𝐡𝐦𝐚𝐫𝐤𝐢𝐧𝐠 & 𝐓𝐮𝐧𝐢𝐧𝐠: Evaluated embedding models using RAGAS, selecting a model that yielded a 31% accuracy improvement.
  • Implemented a critic feedback loop and data annotation processes to fine-tune models and enhance translation quality.
  • • 𝐄𝐧𝐠𝐢𝐧𝐞𝐞𝐫𝐞𝐝 𝐬𝐜𝐚𝐥𝐚𝐛𝐥𝐞 𝐝𝐚𝐭𝐚 𝐩𝐢𝐩𝐞𝐥𝐢𝐧𝐞𝐬 & 𝐈𝐧𝐟𝐫𝐚 on Apple’s Spark and Hive ecosystems, building Tableau dashboards for analytics.
  • Managed CI/CD build images and deployment workflows, and executed database migrations to Conductor (Apple’s cloud alternative).
  • • 𝐃𝐞𝐬𝐢𝐠𝐧𝐞𝐝 𝐚 𝐋𝐞𝐠𝐚𝐥 𝐑𝐢𝐬𝐤 𝐀𝐬𝐬𝐞𝐬𝐬𝐦𝐞𝐧𝐭 𝐀𝐠𝐞𝐧𝐭 using Google ADK & FastAPI that conducts adaptive legal questionnaires, classifies risk levels, and requests supporting documentation intelligently, reducing manual review effort by 60%.

Specialist Programmer - L2

Infosys - Apple client · Jul 2024 – Present

  • Engineering Dataguard platform for PII/PHI/PCI redaction and improving ML models with improved latency & accuracy
  • Lead development of a receipt extraction & translation app with production LLM adoption across 5+ internal teams, achieving remarkable benchmark gains.
  • Built Agentic systems for: BRE documentation generation from cobol repo & Legal risk assessment using Google ADK
  • Dataguard: Currently engineering the Dataguard platform to securely process PII, PHI, & PCI information using regex & machine learning algorithms deployed via Triton inference servers.
  • Architected a decoupled entity validation framework utilizing the Strategy design pattern, decreasing false positives by over 40% (ex.
  • Verhoeff check for Aadhaar, Luhn validation for cc), successfully scaled to support multiple new locales & entities.
  • GenAl system for receipt extraction addresses challenges eg. overlaid text & inconsistent orientation, & translation on Apple's internal ticketing tool using LLMs; adopted in production.
  • Enhancing quality & Apple's policy adherence by automating review processes using LLMs to generate a 13-point custom compliance score.
  • Benchmarked embedding models with RAGAS, using best model evaluated with 31% accuracy improvement; implemented critic feedback loop & annotation for translation quality improvement & fine-tune model.
  • Engineered scalable data pipelines & Infra on Apple's Shuri-Hue (Spark+Hive) & built Tableau dashboards for insights.
  • Managed CI/CD pipelines by developing build images & deployment processes; Executed database migration to Conductor (Apple's AWS alternative).
  • Created Documentation Generator, Agentic Al to perform Business Rules Extraction (BRE) from source code, converting complex logic to structured pseudocode & functional docs with LlamaIndex, Gemini, backend for context-aware workflows, concurrency, & parallel execution for high-performance response handling.
  • Designed a Legal Risk Assessment Agent using Google ADK & FastAPI that conducts adaptive legal questionnaires, classifies risk levels, and requests supporting documentation intelligently, reducing manual review effort by 60%.
Outlier AI

Software Developer - AI Trainer

Outlier AI · Feb 2024 – Jul 2024

  • Trained and fine-tuned Large Language Models using RLHF for complex, real-world coding and reasoning tasks.
  • Improved model accuracy and instruction-following through advanced prompt engineering & agent workflows.
  • Trained LLMs using RLHF, improving performance across Python, MySQL, JavaScript, APIs, and system-level reasoning tasks through structured human feedback and evaluation.
  • Implemented tool-sequencing agents, where LLMs selected and invoked the correct ordered sequence of API tools from a predefined library, ensuring deterministic execution and accurate end-to-end task completion.
  • Built Flamingo (NL
  • SQL), an agentic system that converts natural-language queries into optimized SQL using tools and schema-aware agents, executes queries, and returns grounded natural-language responses with relevant data.
  • Applied multi-turn deviation strategies, stress-testing and fine-tuning model behavior to improve consistency, instruction adherence, and robustness under ambiguous or adversarial inputs.
  • Scaled SFT data pipelines, supporting 500+ annotators for high-throughput Supervised Fine-Tuning with strong quality controls and annotation guidelines.
  • Performed prompt failure analysis, identifying reasoning breakdowns and improving response quality using structured analytical and prompt-optimization techniques.
Outlier

Software Developer - AI Trainer

Outlier · Feb 2024 – Jul 2024

  • Trained and fine-tuned Large Language Models using RLHF for complex, real-world coding and reasoning tasks.
  • Improved model accuracy and instruction-following through advanced prompt engineering & agent workflows.
Wall Street Oasis

Financial Research Analyst

Wall Street Oasis · Jan 2024 – Feb 2024

Amazon

Trainee - Amazon ML Summer School

Amazon · Sep 2023 – Oct 2023

Amazon ML Summer School

Machine Learning Intern

Amazon ML Summer School · Sep 2023 – Oct 2023

  • Key ML topics, including deep neural networks, supervised learning, probabilistic graphical models, & unsupervised learning, Reinforcement Learning, etc., through engaging modules delivered by Amazon Scientists.

Proof-of-Work

No completed sessions yet. Anything Amogh finishes shows up here with the evidence behind it.

Proof of Education

Bachelor of Technology, Electronics & Communication Engineering

Indian Institute of Information Technology Allahabad · 2020 — 2024

A Proof CV — one profile, kept current, with the proof attached. Back to the portal