Prashanth Manchala

#heyprashanth

[Prashanth Manchala @heyprashanth]

AI accelerates output, but without oversight, it can scale errors just as quickly as it scales productivity.

[Prashanth Manchala @heyprashanth]

Do good to others. Always.

[Coforge @Coforge_Tech]

AI isn’t a layer we add. It’s the foundation we engineer.

Explore how we’re reshaping the AI playbook for regulated, complex, and outcome-focused organizations: https://t.co/ZFM2RpcOYJ

#EnterpriseAI #GenAI #AIEngineering

[Coforge @Coforge_Tech]

Coforge to acquire @EncoraInc. This is a defining moment—creating a $2.5Bn AI-led engineering powerhouse, setting new benchmarks in Data, Cloud, and Healthcare innovation.

Learn more: https://t.co/F8twx0Mols

[Prashanth Manchala @heyprashanth]

BCG’s annual survey of global c-suite leaders reveals that businesses are doubling down on cost efficiency to stay competitive amid economic uncertainty and global challenges.

86% executives plan for investing in AI and advanced analytics for efficiency
#bcg #ai #analytics

[Coforge @Coforge_Tech]

29.8x cheaper. 95% lower API costs. 90% accuracy at 2x speed.

#Deepseek cuts memory needs by 75%, processes data in phrases, and leverages reinforced learning to redefine AI.

Explore the future of AI with our AI/ML expert, Deepak Bagchi: https://t.co/BZGU9yKrlx

#GenerativeAI

[Prashanth Manchala @heyprashanth]

What sets Deepseek apart is its R1 model which uses reinforcement learning. Using reinforcement leaning autonomous agent learns to perform a task through trial and error and without any instructions from a human user. #deepseek #reinforcementlearning #ai

[Prashanth Manchala @heyprashanth]

It took Deepseek nearly $6m to built a chatbot using the combination of lower end chips and higher end, challenging the need for bigger budgets and top-tier chips for advancing in AI. #deepseek #ai

[Prashanth Manchala @heyprashanth]

Deepseek’s efficiency stems from innovative multi-stage training and large-scale reinforcement learning. #ai #genai #nvidia #meta #microsoft

[Prashanth Manchala @heyprashanth]

The AI Stack:
1) Semiconductors - #Nvidia, #AMD, #intel
2) Cloud Infrastructures - #aws, #googlecloud #azure #snowflake
3) Foundation Models - #openai #meta #anthropic
The real value is in building applications using the above AI Stack.

[Prashanth Manchala @heyprashanth]

Global growth for the past 30 years was at 4% and the predicted global growth for the year 2025 is 3.3%. But all eyes should be on sustainable growth rather than just the numbers. #WEF2025

[Prashanth Manchala @heyprashanth]

While implementing GenAI based chatbots, businesses leaders should:
1. Prioritize customer experience over cost savings
2. Focus on providing safe AI interactions
3. Manage transaction workflows to deliver positive customer interactions

[Prashanth Manchala @heyprashanth]

The 5 milestones of AI continuum:
1. RPA
2. GenAI
3. Agentic AI
4. Artificial General Intelligence
5. Artificial Super Intelligence

[Prashanth Manchala @heyprashanth]

There is a lot of innovation happening at the service provider level in GenAI space that CIOs may often feel overwhelmed. The right strategy is to find a use case that is best suited for your organization, show the ROI to get the buy-in of the board for further investments.

[Prashanth Manchala @heyprashanth]

If you have modest AI ambitions, that isn’t being remastered by AI yet, you can afford to go at a more measured pace. This is an AI-steady pace. Organizations with bigger AI ambitions, or that’s being reinvented by AI, the pace will be faster. This is an AI-accelerated pace.

[Prashanth Manchala @heyprashanth]

Human interactions with GenAI may evolve from users prompting (LLMs) to users interacting directly with autonomous agents, allowing for a higher level of alignment of humans and machines.

[Prashanth Manchala @heyprashanth]

The global human population crossed 8 billion today.
It has doubled from 4 billion in a span of 48 years. The global population was 4 billion in 1974.#globalpopulation #UnitedNations #datapoints

[Prashanth Manchala @heyprashanth]

85% of the ML models fail to produce desired results. One of the remarkable reasons is not having 'clean data'.
No matter how good the ML model is, if it isn't clean enough, it ought to fail.
#DataCleaning #DataScientists #dataengineering #MLmodel

[Prashanth Manchala @heyprashanth]

A 10% improvement in customer data quality can be linked to a 5% improvement in customer responsiveness. #dataquality #dataengineering

[Morning Brew ☕️ @MorningBrew]

This sweater developed by the University of Maryland is an invisibility cloak against AI.

It uses "adversarial patterns" to stop AI from recognizing the person wearing it.