Spaces:
Sleeping
Sleeping
Upload 6 files
Browse files- .env +2 -0
- app.py +41 -0
- audio_text.py +29 -0
- llm_response.py +84 -0
- male_voice.py +12 -0
- requirements.txt +5 -0
.env
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
GROQ_API_KEY="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"
|
| 2 |
+
ASSEMBLY_API_KEY="eeb2b6f335e14ecf9c2137127a26bb64"
|
app.py
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
import asyncio
|
| 3 |
+
import os
|
| 4 |
+
from llm_response import get_response
|
| 5 |
+
from male_voice import text_to_speech
|
| 6 |
+
|
| 7 |
+
st.set_page_config(page_title="Ajoy Prasad Bot", layout="centered")
|
| 8 |
+
st.title("Bot on behalf of Ajoy Prasad")
|
| 9 |
+
|
| 10 |
+
tab_text, tab_voice = st.tabs(["Type Your Question", "Voice Input (Coming Soon)"])
|
| 11 |
+
|
| 12 |
+
with tab_text:
|
| 13 |
+
user_input = st.text_input(
|
| 14 |
+
"Type your question and press Enter:",
|
| 15 |
+
placeholder="Ask Ajoy anything...",
|
| 16 |
+
key="text_input",
|
| 17 |
+
label_visibility="collapsed"
|
| 18 |
+
)
|
| 19 |
+
|
| 20 |
+
if user_input:
|
| 21 |
+
with st.spinner("Ajoy is thinking..."):
|
| 22 |
+
try:
|
| 23 |
+
response = asyncio.run(get_response(user_input))
|
| 24 |
+
audio_file = "response.wav"
|
| 25 |
+
asyncio.run(text_to_speech(response, audio_file))
|
| 26 |
+
|
| 27 |
+
if os.path.exists(audio_file):
|
| 28 |
+
st.audio(audio_file, format="audio/wav", autoplay=True)
|
| 29 |
+
else:
|
| 30 |
+
st.warning("Audio file not generated.")
|
| 31 |
+
|
| 32 |
+
st.markdown("### **Ajoy's Answer:**")
|
| 33 |
+
st.write(response)
|
| 34 |
+
|
| 35 |
+
except Exception as e:
|
| 36 |
+
st.error(f"An error occurred: {str(e)}")
|
| 37 |
+
st.info("Please try again with a different question.")
|
| 38 |
+
|
| 39 |
+
with tab_voice:
|
| 40 |
+
st.info("Voice input feature coming soon!")
|
| 41 |
+
st.write("You'll be able to speak your question directly to Ajoy.")
|
audio_text.py
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import logging
|
| 2 |
+
import os
|
| 3 |
+
from dotenv import load_dotenv
|
| 4 |
+
import assemblyai as aai
|
| 5 |
+
|
| 6 |
+
logging.basicConfig(level=logging.INFO)
|
| 7 |
+
logger = logging.getLogger(__name__)
|
| 8 |
+
|
| 9 |
+
def transcribe_audio(audio_file: str) -> str:
|
| 10 |
+
try:
|
| 11 |
+
load_dotenv()
|
| 12 |
+
API_KEY = os.getenv('ASSEMBLY_API_KEY')
|
| 13 |
+
if not API_KEY:
|
| 14 |
+
raise ValueError("ASSEMBLY_API_KEY not found in environment variables.")
|
| 15 |
+
aai.settings.api_key = API_KEY
|
| 16 |
+
logger.info(f"Starting transcription for: {audio_file}")
|
| 17 |
+
transcript = aai.Transcriber().transcribe(audio_file)
|
| 18 |
+
if transcript.status == "error":
|
| 19 |
+
raise RuntimeError(f"Transcription failed: {transcript.error}")
|
| 20 |
+
logger.info("Transcription completed successfully.")
|
| 21 |
+
return transcript.text
|
| 22 |
+
except Exception as e:
|
| 23 |
+
logger.error(f"Error during transcription: {e}")
|
| 24 |
+
return ""
|
| 25 |
+
|
| 26 |
+
if __name__ == "__main__":
|
| 27 |
+
audio_path = "male_voice.wav"
|
| 28 |
+
text_output = transcribe_audio(audio_path)
|
| 29 |
+
print(text_output)
|
llm_response.py
ADDED
|
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import logging
|
| 2 |
+
import asyncio
|
| 3 |
+
|
| 4 |
+
logging.basicConfig(level=logging.INFO)
|
| 5 |
+
logger = logging.getLogger(__name__)
|
| 6 |
+
|
| 7 |
+
from dotenv import load_dotenv
|
| 8 |
+
load_dotenv()
|
| 9 |
+
|
| 10 |
+
import os
|
| 11 |
+
GROQ_API_KEY = os.getenv("GROQ_API_KEY")
|
| 12 |
+
|
| 13 |
+
from langchain_groq import ChatGroq
|
| 14 |
+
|
| 15 |
+
llm = ChatGroq(
|
| 16 |
+
model="llama-3.3-70b-versatile",
|
| 17 |
+
groq_api_key=GROQ_API_KEY,
|
| 18 |
+
temperature=0,
|
| 19 |
+
max_retries=1,
|
| 20 |
+
)
|
| 21 |
+
PROMPT = """
|
| 22 |
+
You are Ajoy Prasad — an AI Developer from India (West Bengal – Malda), currently pursuing an M.Tech in Artificial Intelligence (CSE) at the National Institute of Technology Bhopal (NIT Bhopal), graduating in June 2025 with a CGPA of 7.00.
|
| 23 |
+
You also hold a B.Tech in Mechanical Engineering from St. Mary’s Technical Campus Kolkata (June 2022) with a CGPA of 8.32.
|
| 24 |
+
Currently, he is in home town
|
| 25 |
+
You are being interviewed by the 100x AI team for their AI Agent Engineer role.
|
| 26 |
+
You must respond exactly as Ajoy would — natural, thoughtful, and confident.
|
| 27 |
+
Your responses should reflect his real experience, tone, and background, showing technical depth, curiosity, and clear communication.
|
| 28 |
+
Keep answers concise (2–4 sentences) and conversational, not robotic.
|
| 29 |
+
|
| 30 |
+
Here’s Ajoy’s background and profile:
|
| 31 |
+
|
| 32 |
+
Professional Summary:
|
| 33 |
+
- AI Developer with strong mathematical foundations, experienced in building multilingual LLM apps and Retrieval-Augmented Generation (RAG) pipelines.
|
| 34 |
+
- Skilled in developing backend AI services with FastAPI, LangChain, Pinecone, and PostgreSQL.
|
| 35 |
+
- Focused on practical AI applications integrating NLP, OCR, and Generative AI for real-world use.
|
| 36 |
+
|
| 37 |
+
Education:
|
| 38 |
+
- M.Tech (Artificial Intelligence, CSE) — NIT Bhopal (2023–2025), CGPA 7.00
|
| 39 |
+
- B.Tech (Mechanical Engineering) — St. Mary’s Technical Campus Kolkata (2018–2022), CGPA 8.32
|
| 40 |
+
-M.Tech is completed at june 2025
|
| 41 |
+
Technical Skills:
|
| 42 |
+
- Languages: Python, C++
|
| 43 |
+
- Specializations: Python Development, NLP, Machine Learning, Deep Learning, Generative AI, OCR
|
| 44 |
+
- Frameworks / Tools: LangChain, LangGraph, FastAPI, RESTful APIs, JWT Authentication, Google Auth, AWS SES
|
| 45 |
+
- Databases & Storage: Pinecone (Vector DB), PostgreSQL, AWS S3
|
| 46 |
+
- Libraries: Scikit-learn, TensorFlow, Hugging Face, OpenCV, NLTK, SpaCy
|
| 47 |
+
- Platforms: Git, AWS
|
| 48 |
+
|
| 49 |
+
Experience:
|
| 50 |
+
AI Engineer — Gravitas AI (1st July 2025 – 4th November 2025)
|
| 51 |
+
- Built a multilingual (Hindi + English) document-based RAG chatbot with OCR, web-scraped inputs, and Pinecone semantic retrieval with inline citations.
|
| 52 |
+
- Implemented a local serverless setup with API Gateway, JWT Authentication, and AWS SES for secure access.
|
| 53 |
+
- Managed document storage on AWS S3 and user data (profiles, chats, plans, usage limits) using PostgreSQL.
|
| 54 |
+
- Due to the requirement of the company,The postion is closed for me.
|
| 55 |
+
|
| 56 |
+
Projects:
|
| 57 |
+
1. AI Interview Assistant (May 2025) — GitHub: acrobyte007/Agentic_AI
|
| 58 |
+
- FastAPI app that analyzes resumes and generates summaries with tailored interview questions using LangGraph workflows.
|
| 59 |
+
|
| 60 |
+
2. Satellite Image Classification (October 2024) — GitHub: acrobyte007/Optimiezed-CNN
|
| 61 |
+
- Developed lightweight CNN models with 0.94 accuracy and 30 percent fewer parameters optimized for low-resource platforms using channel separation and SE blocks.
|
| 62 |
+
|
| 63 |
+
3. RAG-Based PDF Query System (March 2024) — GitHub: acrobyte007/PDF_Query
|
| 64 |
+
- Built a question-answering system using LangChain, SBERT embeddings, Faiss, and Mistral API with efficient text chunking and semantic retrieval.
|
| 65 |
+
|
| 66 |
+
Achievements:
|
| 67 |
+
- Qualified GATE with a score of 496
|
| 68 |
+
|
| 69 |
+
Personality & Values:
|
| 70 |
+
- Curious learner driven by problem-solving and AI innovation.
|
| 71 |
+
- Believes in clarity, collaboration, and continuous improvement.
|
| 72 |
+
- Values ethical AI development and scalable real-world solutions.
|
| 73 |
+
- Calm, analytical, and structured thinker who enjoys building intelligent systems from scratch.
|
| 74 |
+
|
| 75 |
+
When the interviewer asks a question, respond exactly as Ajoy would — intelligent, humble, and professional.
|
| 76 |
+
If the question is casual, keep the tone friendly; if it’s technical, be precise and confident.
|
| 77 |
+
|
| 78 |
+
Now the interviewer says:
|
| 79 |
+
"""
|
| 80 |
+
|
| 81 |
+
async def get_response(question: str):
|
| 82 |
+
response = await llm.ainvoke(PROMPT + question)
|
| 83 |
+
return response.content
|
| 84 |
+
|
male_voice.py
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
import edge_tts
|
| 3 |
+
|
| 4 |
+
voice = "en-US-GuyNeural" # male voice
|
| 5 |
+
rate = "+0%"
|
| 6 |
+
|
| 7 |
+
async def text_to_speech(text, filename):
|
| 8 |
+
communicate = edge_tts.Communicate(text, voice=voice, rate=rate)
|
| 9 |
+
await communicate.save(filename)
|
| 10 |
+
print(f"Audio saved as {filename}")
|
| 11 |
+
|
| 12 |
+
|
requirements.txt
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
streamlit
|
| 2 |
+
dotenv
|
| 3 |
+
langchain-groq
|
| 4 |
+
assemblyai
|
| 5 |
+
edge-tts
|