Devansh
109 posts
Devansh retweetet
Interestingly, the most common error in agentic systems is the OOM error :)
When you build an agent, almost everything you query and everything it needs is held in memory. Things usually work fine locally, but when you put it into production, or just add a little more load - longer conversations, more tool results, bigger documents - the process dies from running out of memory.
To be honest, these systems work perfectly during development, only to fail once conversations get longer or traffic starts increasing. Of course, the issue is not the model here, but the amount of context and state we keep around it. Here are some common culprits:
- tool results held in context
- search results, API responses, and outputs pile up
- no context truncation in place
- external call results accumulate in memory
- entire doc loaded when only a small portion is needed
Also, the fixes are not exotic in any way: truncate aggressively, summarize, intermediate steps, remove what's not needed from context and mem, retrieve only what is needed, stream instead of buffering, and persist state externally instead of keeping everything in memory.
The key is identifying what truly needs to stay in memory and what can be moved elsewhere. Also, it is equally important to understand how memory usage grows as conversations get longer or agent loops run far beyond what was originally expected.
Given how long-running AI systems are, it is important to design them to handle errors, rate limits, and failure scenarios gracefully, with a well-defined degradation policy.
We have all solved these kinds of problems before. Agentic systems are no different - the constraints are familiar, but the surface area is new.
Hope this helps.
English
At @takeUforward_, we make sure everyone has lunch together.
People from across the team sit together, have fun, share their tiffins, and simply enjoy those 45–60 minutes together ❤️
English
Devansh retweetet
🎁 SDE Sheet Challenge – Goodies Breakdown
Consistency deserves rewards.
We're selecting 50 winners from the SDE Sheet Challenge:
🏆 20 Winners from LinkedIn
🏆 15 Winners from X (Twitter)
🏆 15 Winners from Instagram
How to increase your chances?
→ Solve the daily problems consistently
→ Share your learnings publicly throughout the challenge
→ Be creative with your posts (especially Instagram Reels)
→ Use the official challenge hashtag
→ Tag us in your posts so we can track your participation
Start Date: June 1
Let's see who has the discipline to finish what they start.
Join here to participate whatsapp.com/channel/0029Vb…
English
career update 🎊
sde intern in Chennai at a US based startup
Ayush 🦀@ayushcodesweb
career update loading....⏳
English
my college is such a bitch.
only reason i’m tolerating this shit is because the people make it liveable
English
Interested to join an #internship offering ₹70,000- ₹1,00,000 stipend per month at #Bangalore ?
English
Devansh retweetet
