↓ Skip to main content
Tech Blog
Tech Blog

About

·467 words·3 mins
Yash Sachdeva
Author
Yash Sachdeva
Software Engineer | Turning Complex Problems into Simple Solutions
Software engineer by trade, problem solver by nature. I write about the systems I build in my free time and the experiences that shape them.
Table of Contents

Here’s what I’m focused on right now, and a history of what I’ve built.

📄 Download Resume (PDF)

What I’m doing now
#

Currently building the future of AI at Salesforce as an SDE3 / SMTS. Since late 2025, my primary focus has been designing an Agentic Metadata Generation and Evaluation Framework for Agentforce Vibes. This empowers low-code users to reliably generate metadata through natural language, backed by a holistic, end-to-end evaluation system.

Experience
#

Salesforce
#

SMTS/SDE3 (India) | May 2020 – Present

Custom Apps Gen AI (Jan’25 – Sep’25)

  • Optimized the LLM-based app generation framework via batching, parallelization, and eventual consistency, reducing build time by 25%.
  • Early adopter of Cursor and MCPs within the team; built various internal tooling MCPs (like Sourcegraph) to boost developer productivity and automate Trust efforts.

Commerce Generative AI (Jan’23 – Dec’24)

  • Built and demo’ed a product recommendations prototype using Python that integrated LLMs for B2B/B2C e-commerce with text and image prompts. Led the project to production on a Java/Spring tech stack, leading teams in Brazil and the US.
  • Developed the prototype integrated with large compute AI models on Heroku, which became a key company strategy with projected revenues exceeding $1B.
  • Led global teams to design the first LLM-based features in Salesforce Commerce (Product Recommendations, Conversational Add To Cart, and Merchant Insights).

Buyer And Entitlements (Jan’22 – Dec'22)

  • Led API performance improvements, increasing guardrails by 10x for buyer accounts to onboard large F&B customers by designing a predictive caching framework.
  • Designed multi-tier Redis-based cache systems supporting multiple lookups with a p95 time of <50ms.

Address Management & Site Search (Jun’20 – Dec’21)

  • Led development of Address APIs driving formats for 4076+ enterprise customers (FedEx, Louis Vuitton).
  • Contributed to Salesforce search platform code reducing GTM time for new search products by 70%. Built custom monitoring dashboards, cutting incident response time from 1 hour to near-instant.

Arcesium
#

Senior Software Engineer | Nov 2017 – May 2020

  • Revamped pricing models to reduce manual effort by 90%, facilitating the onboarding of 3 clients with $100M portfolios.
  • Optimized quote-fetching logic to save $1M+ for 20+ clients and built bulk APIs reducing data entry time by 99% (3 hours to 45 secs).

Wooqer
#

Platform Engineer | Jul 2017 – Nov 2017

  • Developed iOS features that reduced long-term maintenance costs by 50%.

Technical Skills
#

  • Languages: Java (Proficient); Python, Javascript, C, C++
  • Databases: SQL Server, Oracle, PostgreSQL, Redis
  • Architecture & AI: Distributed Systems, System Design, Evals, RAG, MCPs, Vector Databases, LangChain, Hugging Face
  • Tools: IntelliJ IDEA, Git, Claude CLI, Cursor, Antigravity, LangSmith, Splunk, Grafana

Education & Extracurriculars
#

  • BTech - Computer Science Engineering, IIIT Kota, India (CGPA: 8.99/10, Rank 3)
  • Extracurriculars:
    • Salesforce TMP Spring’23 Cloud Hackathon winner (Global).
    • Salesforce Commerce Cloud Summer Hackathon winner.
    • Career Coach at InterviewBit (‘20 – ’24).
    • Yoga/Crossfit enthusiast, 100km Cyclothon finisher.